-
Notifications
You must be signed in to change notification settings - Fork 1.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Fix --gen_kwargs
and VLLM (temperature
not respected)
#1800
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
lintangsutawika
approved these changes
Jun 11, 2024
Mogreine
pushed a commit
to deepvk/lm-evaluation-harness
that referenced
this pull request
Jun 25, 2024
* Update generate_until_template_yaml (EleutherAI#1546) * Update ifeval.yaml (EleutherAI#1506) * add Arabic EXAMS benchmark (EleutherAI#1498) * add Arabic EXAMS benchmark * fixed the linter issue, and add more information on the readme * Update README.md --------- Co-authored-by: Lintang Sutawika <[email protected]> * AGIEval (EleutherAI#1359) * add agieval * fix typo * add cloze / math exactmatch agieval tasks, rename * update exact-match agieval tasks, allow for multiple-correct answers * add more detail to readme * don't parse_math_answer twice --------- Co-authored-by: Alex Bäuerle <[email protected]> * cli_evaluate calls simple_evaluate with the same verbosity. (EleutherAI#1563) * add manual tqdm disabling management (EleutherAI#1569) * add manual tqdm disabling management * add typing to all new args * apply precommit changes --------- Co-authored-by: haileyschoelkopf <[email protected]> * Fix README section on vllm integration (EleutherAI#1579) * Link to vllm integration * add pip install .[vllm] cmd * Fix Jinja template for Advanced AI Risk (EleutherAI#1587) * Proposed approach for testing CLI arg parsing (EleutherAI#1566) * New tests for CLI args * fix spacing * change tests for parsing * add tests, fix parser * remove defaults for store_true * Patch for Seq2Seq Model predictions (EleutherAI#1584) * Differentiate _encode_pair setting for decoder and enc-dec models * tok_decode to not skip special token so that eos doen't become empty string * Update model.py * Update model.py * Update huggingface.py * Update lm_eval/models/huggingface.py Co-authored-by: Hailey Schoelkopf <[email protected]> * Update model.py --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Add start date in results.json (EleutherAI#1592) * Cleanup for v0.4.2 release (EleutherAI#1573) * Update interface.md * fix: make caching reqs always work with accelerate launch * remove stale task migration checklist * remove deprecation warnings * make informative TypeErrors for get_task_dict * bump version metadata * fix num_fewshot printing bug * add fewshot value to cache key * Fix eval_logger import for mmlu/_generate_configs.py (EleutherAI#1593) * Fix eval_logger import for mmlu/_generate_configs.py * linter --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * use BOS token in loglikelihood (EleutherAI#1588) * use BOS token in loglikelihood * improve comments * add model arg * log prefix token id * log prefix token id * Update lm_eval/api/model.py Co-authored-by: Hailey Schoelkopf <[email protected]> * change name to prefix_token_id --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Revert "Patch for Seq2Seq Model predictions (EleutherAI#1584)" (EleutherAI#1601) This reverts commit b7923a8. * fix gen_kwargs arg reading (EleutherAI#1607) * fix until arg processing (EleutherAI#1608) * Fixes to Loglikelihood prefix token / VLLM (EleutherAI#1611) * make vllm use prefix_token_id ; have prefix_token_id be optional method to define * custom_prefix_token_id wasn't set if not passed * Add ACLUE task (EleutherAI#1614) * Add task ACLUE * fix minor bug * fix code style * fix code style * OpenAI Completions -- fix passing of unexpected 'until' arg (EleutherAI#1612) * add logging of model args (EleutherAI#1619) * add logging of model args * nit * Add warnings. * nit * add warning * nit * Add vLLM FAQs to README (EleutherAI#1625) (EleutherAI#1633) * peft Version Assertion (EleutherAI#1635) * peft Version Assertion * fix the linter issue * Seq2seq fix (EleutherAI#1604) * fix on --task list * add fixes to tokeniation * differentiate encoding for seq2seq and decoder * return token setting * format for pre-commit * Seq2seq fix, pt2 (EleutherAI#1630) * getting model class only when defined * encode_pair handles None, add_special_tokens turned into dict with default value --------- Co-authored-by: achervyakov <[email protected]> * Integration of NeMo models into LM Evaluation Harness library (EleutherAI#1598) * Integration of NeMo models into LM Evaluation Harness library * rename nemo model as nemo_lm * move nemo section in readme after hf section * use self.eot_token_id in get_until() * improve progress bar showing loglikelihood requests * data replication or tensor/pipeline replication working fine within one node * run pre-commit on modified files * check whether dependencies are installed * clarify usage of torchrun in README * Fix conditional import for Nemo LM class (EleutherAI#1641) * Fix SuperGlue's ReCoRD task following regression in v0.4 refactoring (EleutherAI#1647) * Add Latxa paper evaluation tasks for Basque (EleutherAI#1654) * add basqueglue * add eus_exams * add eus_proficiency * add eus_reading * add eus_trivia * run pre-commit * Fix CLI --batch_size arg for openai-completions/local-completions (EleutherAI#1656) The OpenAI interface supports batch size as an argument to the completions API, but does not seem to support specification of this on the CLI i.e. `lm_eval --model openai-completions --batch_size 16 ...` because of a simple lack of str->int conversion. This is confirmed by my usage and stacktrace from running `OPENAI_API_KEY=dummy lm_eval --model local-completions --tasks gsm8k --batch_size 16 --model_args model=nm- testing/zephyr-beta-7b-gptq-g128,tokenizer_backend=huggingface,base_url=http:https://localhost:8000/v1`: ``` Traceback (most recent call last): File "/home/michael/venv/bin/lm_eval", line 8, in <module> sys.exit(cli_evaluate()) File "/home/michael/code/lm-evaluation-harness/lm_eval/__main__.py", line 341, in cli_evaluate results = evaluator.simple_evaluate( File "/home/michael/code/lm-evaluation-harness/lm_eval/utils.py", line 288, in _wrapper return fn(*args, **kwargs) File "/home/michael/code/lm-evaluation-harness/lm_eval/evaluator.py", line 251, in simple_evaluate results = evaluate( File "/home/michael/code/lm-evaluation-harness/lm_eval/utils.py", line 288, in _wrapper return fn(*args, **kwargs) File "/home/michael/code/lm-evaluation-harness/lm_eval/evaluator.py", line 390, in evaluate resps = getattr(lm, reqtype)(cloned_reqs) File "/home/michael/code/lm-evaluation-harness/lm_eval/models/openai_completions.py", line 263, in generate_until list(sameuntil_chunks(re_ord.get_reordered(), self.batch_size)), File "/home/michael/code/lm-evaluation-harness/lm_eval/models/openai_completions.py", line 251, in sameuntil_chunks if len(ret) >= size or x[1] != lastuntil: TypeError: '>=' not supported between instances of 'int' and 'str' ``` * Patch QQP prompt (EleutherAI#1661) * TMMLU+ implementation (EleutherAI#1394) * implementation of TMMLU+ * implemented: TMMLU+ ****TMMLU+ : large-scale Traditional chinese Massive Multitask language Understanding**** - 4 categories - STEM - Social Science - Humanities - Other The TMMLU+ dataset, encompassing over 67 subjects and 20160 tasks, is six times larger and more balanced than its predecessor, TMMLU, and includes benchmark results from both closed-source and 20 open-weight Chinese large language models with 1.8B to 72B parameters. However, Traditional Chinese variants continue to underperform compared to major Simplified Chinese models. ```markdown Total number of tasks in the 'test' sets: 20160 Total number of tasks in the 'validation' sets: 2247 Total number of tasks in the 'train' sets: 335 ``` * Remove print from __init__.py There was my mistake in forgetting to remove the debug print from the code. * update: move TMMLU+ config generation program into default * fix: we should use training set as few shots example * update: README for TMMLU+ * update: a small changes of TMMLU+ README file * pre-commit run thought * Add README for TMMLU+ dataset * run precommit * trigger precommit again * trigger precommit again * isort is fussy * isort is fussy * format, again * oops * oops --------- Co-authored-by: lintang <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Anthropic Chat API (EleutherAI#1594) * claude3 * supply for anthropic claude3 * supply for anthropic claude3 * anthropic config changes * add callback options on anthropic * line passed * claude3 tiny change * help anthropic installation * mention sysprompt / being careful with format in readme --------- Co-authored-by: haileyschoelkopf <[email protected]> * correction bug EleutherAI#1664 (EleutherAI#1670) * correction bug EleutherAI#1664 * add any invalid characters for Windows filenames and Unix-like systems see: https://gist.github.com/doctaphred/d01d05291546186941e1b7ddc02034d3?permalink_comment_id=3958715 * Update lm_eval/__main__.py * Update scripts/zeno_visualize.py * fix format --------- Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Update README.md (EleutherAI#1680) * Add delta weights model loading (EleutherAI#1712) * added delta weights * removed debug * readme update * better error handling * autogptq warn * warn update * peft and delta error, explicitly deleting _model_delta * linter fix * Add `neuralmagic` models for `sparseml` and `deepsparse` (EleutherAI#1674) * Add neuralmagic models for SparseML and DeepSparse * Update to latest and add test * Format * Fix list to List * Format * Add deepsparse/sparseml to automated testing * Update pyproject.toml * Update pyproject.toml * Update README * Fixes for dtype and device * Format * Fix test * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> * Address review comments! --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * fix error when appending eot_token_id for generate_until tasks (EleutherAI#1699) * Adding retries and rate limit to toxicity tasks (EleutherAI#1620) * reference `--tasks list` in README (EleutherAI#1726) EleutherAI#1698 * Add XNLIeu: a dataset for cross-lingual NLI in Basque (EleutherAI#1694) * add xnli_eu tasks * update tasks readme * update readme * Fix Parameter Propagation for Tasks that have `include` (EleutherAI#1749) * Update task.py * Update __init__.py * Support individual scrolls datasets (EleutherAI#1740) * Support individual scrolls datasets * Add qmsum context * Fix formatting * Add filter registry decorator (EleutherAI#1750) * Add register_filter decorator * Add register_filter docs * remove duplicated `num_fewshot: 0` (EleutherAI#1769) * Pile 10k new task (EleutherAI#1758) * Add Pile-10k readme * Add Pile-10k task configuration file * Fix m_arc choices (EleutherAI#1760) * Update utils.py This is a 4-choice task, option_e is null for all but 3 samples * Fix options Adaptive choices * add option e * bump multilingual arc version --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * upload new tasks (EleutherAI#1728) * upload new tasks * add readmes * run linters --------- Co-authored-by: haileyschoelkopf <[email protected]> * vllm lora support (EleutherAI#1756) * vllm lora support * remove print * version check, rename lora kwarg * Add option to set OpenVINO config (EleutherAI#1730) * Add option to set OpenVINO config * Use utils.eval_logger for logging * evaluation tracker implementation (EleutherAI#1766) * evaluation tracker implementation * OVModelForCausalLM test fix * typo fix * moved methods args * multiple args in one flag * loggers moved to dedicated dir * improved filename sanitization * eval tracker args fix (EleutherAI#1777) * limit fix (EleutherAI#1785) * remove echo parameter in OpenAI completions API (EleutherAI#1779) * remove echo parameter in OpenAI completions API * remove context length parameter doc string * Fix README: change`----hf_hub_log_args` to `--hf_hub_log_args` (EleutherAI#1776) fix `----hf_hub_log_args` to `--hf_hub_log_args` * Fix bug in setting until kwarg in openai completions (EleutherAI#1784) * Provide ability for custom sampler for ConfigurableTask (EleutherAI#1616) * Added fewshot sampling seeds to evaluator.simple_evaluate signature Way to control seed of fewshot sampling may help with EleutherAI#1591 * Added ability for custom sampler for ConfigurableTask May be set in config like ``` fewshot_config: sampler: !function utils.MyFewshotSampler ``` * explicitly set fewshot random generator seed for HFLM generate_until_task test * add backward compatibility for three args seed setup * save seeds info to logs/reports * Update `--tasks list` option in interface documentation (EleutherAI#1792) * Fix Caching Tests ; Remove `pretrained=gpt2` default (EleutherAI#1775) * link to the example output on the hub (EleutherAI#1798) * Re-add Hendrycks MATH (no sympy checking, no Minerva hardcoded prompt) variant (EleutherAI#1793) * add Hendrycks MATH (no sympy checking) variant * add readmes for MATH tasks * Logging Updates (Alphabetize table printouts, fix eval tracker bug) (EleutherAI#1774) (EleutherAI#1791) * fix auto-batch size bug for seq2seq models * alphabetize task + group tables ; fix eval tracker bug * fix eval tracker bug * Initial integration of the Unitxt to LM eval harness (EleutherAI#1615) * Initial support for Unitxt datasets in LM Eval Harness See https://github.com/IBM/unitxt The script 'generate_yamls.py' creates LM Eval Harness yaml files corresponding to Unitxt datasets specified in the 'unitxt_datasets' file. The glue code required to register Unitxt metrics is in 'unitxt_wrapper.py'. * Added dataset loading check to generate_yaml Improved error messages. * Speed up generate_yaml Added printouts and improved error message * Added output printout * Simplified integration of unitxt datasets Store all the common yaml configuration in a yaml include shared by all datasets of the same task. * Post code review comments - part 1 1. Made sure include files don't end wth 'yaml' so they won't be marked as tasks 2. Added more datasets and tasks (NER, GEC) 3. Added README * Post code review comments - part 2 1. Added install unitxt install option in pyproject.toml: pip install 'lm_eval[unitxt]' 2. Added a check that unitxt is installed and print a clear error message if not * Commited missing pyproject change * Added documentation on adding datasets * More doc changes * add unitxt extra to readme * run precommit --------- Co-authored-by: haileyschoelkopf <[email protected]> * add task for mmlu evaluation in arc multiple choice format (EleutherAI#1745) * add mmlu arc style evaluation * rename arc_style to continuation --------- Co-authored-by: Jonathan Burdge <[email protected]> Co-authored-by: Jonathan Burdge <[email protected]> * Update flag `--hf_hub_log_args` in interface documentation (EleutherAI#1806) * update interface documentation with flag --hf_hub_logs_arg * update interface documentation with flag --hf_hub_logs_arg 2 * Copal task (EleutherAI#1803) * add copal * change name to copal id for clarity and the task name * remove `copal_id...` to yaml to make it work * checkmark on README * change group name to `copal_id` * Adding tinyBenchmarks datasets (EleutherAI#1545) * Add tinyBenchmarks * Add acknowledgements * Add ordering of outputs for data-parallel * Run pre-commit * Add few_shot specifications * Add tinyBenchmarks post-processing * add conditional import ; fix task names --------- Co-authored-by: haileyschoelkopf <[email protected]> * interface doc update (EleutherAI#1807) * Fix links in README guiding to another branch (EleutherAI#1838) * Fix: support PEFT/LoRA with added tokens (EleutherAI#1828) * resize model embeddings * resize only * tokenizer help * load tokenizer before model * add comment and run precommit lint * Add log message Co-authored-by: Hailey Schoelkopf <[email protected]> --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * fixed incorrect check for task type (replace `~` with `not`) (EleutherAI#1865) * fixed docs typos (EleutherAI#1863) * Update polemo2_out.yaml (EleutherAI#1871) * Unpin vllm in dependencies (EleutherAI#1874) * Fix outdated links to the latest links in `docs` (EleutherAI#1876) * [HFLM]Use Accelerate's API to reduce hard-coded CUDA code (EleutherAI#1880) * Fix `batch_size=auto` for HF Seq2Seq models (EleutherAI#1765) (EleutherAI#1790) * fix auto-batch size bug for seq2seq models * run linter * Fix Brier Score (EleutherAI#1847) `gold_one_hot` needs to follow the dimension of predictions so that it still works when `--limit` is used and the indexes in gold does not cover all gold indexes. * Fix for bootstrap_iters = 0 case (EleutherAI#1715) (EleutherAI#1789) * add handling for bootstrap_iters=0 case * add more detail to docstring * run precommit * add mmlu tasks from pile-t5 (EleutherAI#1710) * add mmlu tasks from pile-t5 * Update _mmlu_flan_cot_fewshot_template_yaml * Update _mmlu_flan_cot_zeroshot_template_yaml * Update _mmlu_flan_generative_template_yaml * Update _mmlu_flan_loglikelihood_template_yaml * Update _default_template_yaml --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Bigbench fix (EleutherAI#1686) * edit process multiple-choice * split template yaml * remove * modified multiple_choice tasks * udpate * Update multiple_choice_template_b_yaml * Update multiple_choice_template_a_yaml --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Rename `lm_eval.logging -> lm_eval.loggers` (EleutherAI#1858) * rename lm_eval.logging module * fix evaluation tracker args * Updated vllm imports in vllm_causallms.py (EleutherAI#1890) * Reorder vllm imports in vllm_causallms.py * Update vllm_causallms.py * [HFLM]Add support for Ascend NPU (EleutherAI#1886) * [HFLM]Add support for Ascend NPU Co-authored-by: jiaqiw09 <[email protected]> Co-authored-by: zhabuye <[email protected]> * bump accelerate dependency version to 0.26.0 for NPU compat. --------- Co-authored-by: jiaqiw09 <[email protected]> Co-authored-by: zhabuye <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> * `higher_is_better` tickers in output table (EleutherAI#1893) * Higher is better tickers in output table * add extra check for `higher_is_better` not being None already * Update lm_eval/evaluator.py * fixup format I messed up * add comment (and retrigger tests) --------- Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Add dataset card when pushing to HF hub (EleutherAI#1898) * dataset card initial * few fixes * adds groups for math, mmlu, gpqa * added summary agrs * moved sanitize_list to utils * readme update * recreate metadata moved * multiple model support * results latest split fix * readme update and small refactor * fix grouping * add comments * added pathlib * corrected pathlib approach * check whether to create a metadata card * convert posix paths to str * default hf org from token * hf token value error * Add logs after successful upload * logging updates * dataset card example in the readme --------- Co-authored-by: Nathan Habib <[email protected]> Co-authored-by: Alina Lozovskaia <[email protected]> * Making hardcoded few shots compatible with the chat template mechanism (EleutherAI#1895) * init test 1 * fix * this format seems to be working - need to update all other tasks with the new format * bbh with few shot format * fix fewshot bbh * add mmlu flan cot * samples of cot * kmmlu * fix gsm8k * update keys for mmlu * minerva math * bbh * fix * fix samples * small fixes to templates * last prompt format change * fixing prompt * fixed minerva math format * rm accidental commited file * added doc for few shot samples * Update lm_eval/loggers/evaluation_tracker.py * Update lm_eval/loggers/evaluation_tracker.py * Update docs/new_task_guide.md Co-authored-by: Hailey Schoelkopf <[email protected]> * added check in sampler per code review * added the system from a function, plus an example in minerva math * style * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> * fix unit tests 1 * forcing use of test split --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Try to make existing tests run little bit faster (EleutherAI#1905) * Fix fewshot seed only set when overriding num_fewshot (EleutherAI#1914) Fix EleutherAI#1906 * Complete task list from pr 1727 (EleutherAI#1901) * added tasks and task family descriptors * continue work on task list w/ links; slightly reorganize README * Apply suggestions from code review * Rename file so that it'll preview in Github when viewing lm_eval/tasks folder * Update new_task_guide.md * Update README.md * run linter * Add language column to task table; Add missing tasks to task table; fix nq_open and storycloze READMEs * fix typo * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> * apply format --------- Co-authored-by: Harish Vadaparty <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Add chat template (EleutherAI#1873) * initial chat template * tokenizer attribute check * variable rename * interface update * system instruction * system inst default update * fewshot as multiturn * typing update * indent update * added comments * Adding a fewshot in a more readable way * linting * Moved apply chat template to LM * multiturn alternation fix * cache key update * apply chat template method fix * add system prompt hash to cache_key * tokenizer name property for cache_key * property name fix * linting backward compatibility fix * docs and errors update * add documentation on adding chat template compatibility to model_guide * fewshot as multiturn check fix * saving system inst and chat template in results * eval tracker update * docs update * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> --------- Co-authored-by: haileyschoelkopf <[email protected]> Co-authored-by: Clémentine Fourrier <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> * Multiple Choice Questions and Large Languages Models: A Case Study with Fictional Medical Data (EleutherAI#1867) * glianorex tasks * Create README.md * Update README.md * Update README.md * fix formatting * fix internal formatting * Modify pre-commit hook to check merge conflicts accidentally committed not at current merge commit (EleutherAI#1927) * [add] fld logical formula task (EleutherAI#1931) * Add new Lambada translations (EleutherAI#1897) * added tasks and task family descriptors * configs for the new lambada translations * continue work on task list w/ links; slightly reorganize README * Apply suggestions from code review * Rename file so that it'll preview in Github when viewing lm_eval/tasks folder * Update new_task_guide.md * Update README.md * run linter * Add language column to task table; Add missing tasks to task table; fix nq_open and storycloze READMEs * fix typo * update `lm_eval/tasks/README.md` with task description --------- Co-authored-by: Harish Vadaparty <[email protected]> Co-authored-by: anthony <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Implement NoticIA (EleutherAI#1912) * Noticia * test * Final testes implementation * Fixes * Fix linters * Add The Arabic version of the PICA benchmark (EleutherAI#1917) * Update siqa.yaml (EleutherAI#1909) * Update basque-glue (EleutherAI#1913) * Update README.md * Update bec.yaml * Update bhtc.yaml * Update coref.yaml * Update qnli.yaml * Update vaxx.yaml * Update wic.yaml * Test output table layout consistency (EleutherAI#1916) * sort metrics in output table * update docstring in `consolidate_results` * add tests for verifying consistency of table output * update tests to account for floating point inconsistencies * updated tests based on `pythia-14m` * Update __main__.py (EleutherAI#1939) * Add the Arabic version with refactor to Arabic pica to be in alghafa folder (EleutherAI#1940) * Results filenames handling fix (EleutherAI#1926) * results filenames handling moved to utils * zeno results handling fix * tasks_for_model backward compatibility * results files logic moved to tasks_for_model * moved sanitize_model_name to utils * Remove AMMLU Due to Translation (EleutherAI#1948) * Update README.md * Delete lm_eval/tasks/ammlu directory * add include_defaults kwarg to taskmanager, add tests for include_path (EleutherAI#1856) * add hacky add_bos_token forcing for Gemma to VLLM too (EleutherAI#1857) * Update interface.md (EleutherAI#1955) * Fix self.max_tokens in anthropic_llms.py (EleutherAI#1848) Fix bug where `self.max_tokens` was not set * `samples` is newline delimited (EleutherAI#1930) * `samples` is newline delimited * updated git and pre-commit * appease pre-commit * nit * Revert back for now * Revert for now --------- Co-authored-by: Lintang Sutawika <[email protected]> * Fix `--gen_kwargs` and VLLM (`temperature` not respected) (EleutherAI#1800) * Update vllm_causallms.py * adjust --------- Co-authored-by: lintangsutawika <[email protected]> * make write_out.py explicitly error if no splits match (EleutherAI#1796) Co-authored-by: lintangsutawika <[email protected]> * fix: add directory filter to os.walk to ignore 'ipynb_checkpoints' (EleutherAI#1956) * fix: add filter to os.walk to ignore 'ipynb_checkpoints * Update __init__.py * Update __init__.py --------- Co-authored-by: Lintang Sutawika <[email protected]> * add trust_remote_code for piqa (EleutherAI#1983) Signed-off-by: changwangss <[email protected]> * Fix self assignment in neuron_optimum.py (EleutherAI#1990) * [New Task] Add Paloma benchmark (EleutherAI#1928) * init paloma benchmark * pre-process in utils function * add `task_alias` * updated task aliases * Update paloma_dolma-v1_5.yaml * Update paloma_twitterAAE_HELM_fixed.yaml * Update paloma_dolma_100_programing_languages.yaml --------- Co-authored-by: Lintang Sutawika <[email protected]> * Fix Paloma Template yaml (EleutherAI#1993) * init paloma benchmark * pre-process in utils function * add `task_alias` * updated task aliases * Update paloma_dolma-v1_5.yaml * Update paloma_twitterAAE_HELM_fixed.yaml * Update paloma_dolma_100_programing_languages.yaml * update on names * fix paloma template issue --------- Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: Lintang Sutawika <[email protected]> * Log `fewshot_as_multiturn` in results files (EleutherAI#1995) * log fewshot_as_multiturn in general tracker args * Update evaluator.py --------- Co-authored-by: Lintang Sutawika <[email protected]> * Added ArabicMMLU (EleutherAI#1987) * Added ArabicMMLU * Rename `ammlu` to `arabicmmlu` * Fix Datasets `--trust_remote_code` (EleutherAI#1998) * Add BertaQA dataset tasks (EleutherAI#1964) * add bertaqa tasks * rename basquetrivia-->bertaqa ; make template stub not .yaml * add bertaqa entry to lm_eval/tasks/README.md --------- Co-authored-by: haileyschoelkopf <[email protected]> * Fix precommit hook, update run_models.sh --------- Signed-off-by: changwangss <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: khalil <[email protected]> Co-authored-by: Lintang Sutawika <[email protected]> Co-authored-by: Alex Bäuerle <[email protected]> Co-authored-by: Wongboo <[email protected]> Co-authored-by: achervyakov <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> Co-authored-by: Eitan Turok <[email protected]> Co-authored-by: Rylan Schaeffer <[email protected]> Co-authored-by: Vicki Boykis <[email protected]> Co-authored-by: Lintang Sutawika <[email protected]> Co-authored-by: kwrobel.eth <[email protected]> Co-authored-by: Nouf M. Alotaibi <[email protected]> Co-authored-by: Haonan Li <[email protected]> Co-authored-by: Baber Abbasi <[email protected]> Co-authored-by: WoosungMyung <[email protected]> Co-authored-by: Sergio Perez <[email protected]> Co-authored-by: Or Sharir <[email protected]> Co-authored-by: Julen Etxaniz <[email protected]> Co-authored-by: Michael Goin <[email protected]> Co-authored-by: ZoneTwelve <[email protected]> Co-authored-by: Seungwoo Ryu <[email protected]> Co-authored-by: nicho2 <[email protected]> Co-authored-by: KonradSzafer <[email protected]> Co-authored-by: Sergio Perez <[email protected]> Co-authored-by: sator-labs <[email protected]> Co-authored-by: Brian Vaughan <[email protected]> Co-authored-by: giorgossideris <[email protected]> Co-authored-by: Nikita Lozhnikov <[email protected]> Co-authored-by: Chujie Zheng <[email protected]> Co-authored-by: Gabriel Mukobi <[email protected]> Co-authored-by: Zehan Li <[email protected]> Co-authored-by: Simran Arora <[email protected]> Co-authored-by: bcicc <[email protected]> Co-authored-by: Helena Kloosterman <[email protected]> Co-authored-by: Muhammad Bin Usman <[email protected]> Co-authored-by: ciaranby <[email protected]> Co-authored-by: LSinev <[email protected]> Co-authored-by: aditya thomas <[email protected]> Co-authored-by: Yoav Katz <[email protected]> Co-authored-by: jonabur <[email protected]> Co-authored-by: Jonathan Burdge <[email protected]> Co-authored-by: Jonathan Burdge <[email protected]> Co-authored-by: Edd <[email protected]> Co-authored-by: Lucas Weber <[email protected]> Co-authored-by: Nick Doiron <[email protected]> Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: zhabuye <[email protected]> Co-authored-by: Edward Gan <[email protected]> Co-authored-by: DongGeon Lee <[email protected]> Co-authored-by: Huazhong Ji <[email protected]> Co-authored-by: jiaqiw09 <[email protected]> Co-authored-by: zhabuye <[email protected]> Co-authored-by: Nathan Habib <[email protected]> Co-authored-by: Alina Lozovskaia <[email protected]> Co-authored-by: Clémentine Fourrier <[email protected]> Co-authored-by: anthony-dipofi <[email protected]> Co-authored-by: Harish Vadaparty <[email protected]> Co-authored-by: Maxime <[email protected]> Co-authored-by: MorishT <[email protected]> Co-authored-by: Iker García-Ferrero <[email protected]> Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: Sadra Barikbin <[email protected]> Co-authored-by: johnwee1 <[email protected]> Co-authored-by: Wang, Chang <[email protected]> Co-authored-by: Yazeed Alnumay <[email protected]>
Mogreine
pushed a commit
to deepvk/lm-evaluation-harness
that referenced
this pull request
Jun 25, 2024
* Update generate_until_template_yaml (EleutherAI#1546) * Update ifeval.yaml (EleutherAI#1506) * add Arabic EXAMS benchmark (EleutherAI#1498) * add Arabic EXAMS benchmark * fixed the linter issue, and add more information on the readme * Update README.md --------- Co-authored-by: Lintang Sutawika <[email protected]> * AGIEval (EleutherAI#1359) * add agieval * fix typo * add cloze / math exactmatch agieval tasks, rename * update exact-match agieval tasks, allow for multiple-correct answers * add more detail to readme * don't parse_math_answer twice --------- Co-authored-by: Alex Bäuerle <[email protected]> * cli_evaluate calls simple_evaluate with the same verbosity. (EleutherAI#1563) * add manual tqdm disabling management (EleutherAI#1569) * add manual tqdm disabling management * add typing to all new args * apply precommit changes --------- Co-authored-by: haileyschoelkopf <[email protected]> * Fix README section on vllm integration (EleutherAI#1579) * Link to vllm integration * add pip install .[vllm] cmd * Fix Jinja template for Advanced AI Risk (EleutherAI#1587) * Proposed approach for testing CLI arg parsing (EleutherAI#1566) * New tests for CLI args * fix spacing * change tests for parsing * add tests, fix parser * remove defaults for store_true * Patch for Seq2Seq Model predictions (EleutherAI#1584) * Differentiate _encode_pair setting for decoder and enc-dec models * tok_decode to not skip special token so that eos doen't become empty string * Update model.py * Update model.py * Update huggingface.py * Update lm_eval/models/huggingface.py Co-authored-by: Hailey Schoelkopf <[email protected]> * Update model.py --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Add start date in results.json (EleutherAI#1592) * Cleanup for v0.4.2 release (EleutherAI#1573) * Update interface.md * fix: make caching reqs always work with accelerate launch * remove stale task migration checklist * remove deprecation warnings * make informative TypeErrors for get_task_dict * bump version metadata * fix num_fewshot printing bug * add fewshot value to cache key * Fix eval_logger import for mmlu/_generate_configs.py (EleutherAI#1593) * Fix eval_logger import for mmlu/_generate_configs.py * linter --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * use BOS token in loglikelihood (EleutherAI#1588) * use BOS token in loglikelihood * improve comments * add model arg * log prefix token id * log prefix token id * Update lm_eval/api/model.py Co-authored-by: Hailey Schoelkopf <[email protected]> * change name to prefix_token_id --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Revert "Patch for Seq2Seq Model predictions (EleutherAI#1584)" (EleutherAI#1601) This reverts commit b7923a8. * fix gen_kwargs arg reading (EleutherAI#1607) * fix until arg processing (EleutherAI#1608) * Fixes to Loglikelihood prefix token / VLLM (EleutherAI#1611) * make vllm use prefix_token_id ; have prefix_token_id be optional method to define * custom_prefix_token_id wasn't set if not passed * Add ACLUE task (EleutherAI#1614) * Add task ACLUE * fix minor bug * fix code style * fix code style * OpenAI Completions -- fix passing of unexpected 'until' arg (EleutherAI#1612) * add logging of model args (EleutherAI#1619) * add logging of model args * nit * Add warnings. * nit * add warning * nit * Add vLLM FAQs to README (EleutherAI#1625) (EleutherAI#1633) * peft Version Assertion (EleutherAI#1635) * peft Version Assertion * fix the linter issue * Seq2seq fix (EleutherAI#1604) * fix on --task list * add fixes to tokeniation * differentiate encoding for seq2seq and decoder * return token setting * format for pre-commit * Seq2seq fix, pt2 (EleutherAI#1630) * getting model class only when defined * encode_pair handles None, add_special_tokens turned into dict with default value --------- Co-authored-by: achervyakov <[email protected]> * Integration of NeMo models into LM Evaluation Harness library (EleutherAI#1598) * Integration of NeMo models into LM Evaluation Harness library * rename nemo model as nemo_lm * move nemo section in readme after hf section * use self.eot_token_id in get_until() * improve progress bar showing loglikelihood requests * data replication or tensor/pipeline replication working fine within one node * run pre-commit on modified files * check whether dependencies are installed * clarify usage of torchrun in README * Fix conditional import for Nemo LM class (EleutherAI#1641) * Fix SuperGlue's ReCoRD task following regression in v0.4 refactoring (EleutherAI#1647) * Add Latxa paper evaluation tasks for Basque (EleutherAI#1654) * add basqueglue * add eus_exams * add eus_proficiency * add eus_reading * add eus_trivia * run pre-commit * Fix CLI --batch_size arg for openai-completions/local-completions (EleutherAI#1656) The OpenAI interface supports batch size as an argument to the completions API, but does not seem to support specification of this on the CLI i.e. `lm_eval --model openai-completions --batch_size 16 ...` because of a simple lack of str->int conversion. This is confirmed by my usage and stacktrace from running `OPENAI_API_KEY=dummy lm_eval --model local-completions --tasks gsm8k --batch_size 16 --model_args model=nm- testing/zephyr-beta-7b-gptq-g128,tokenizer_backend=huggingface,base_url=http:https://localhost:8000/v1`: ``` Traceback (most recent call last): File "/home/michael/venv/bin/lm_eval", line 8, in <module> sys.exit(cli_evaluate()) File "/home/michael/code/lm-evaluation-harness/lm_eval/__main__.py", line 341, in cli_evaluate results = evaluator.simple_evaluate( File "/home/michael/code/lm-evaluation-harness/lm_eval/utils.py", line 288, in _wrapper return fn(*args, **kwargs) File "/home/michael/code/lm-evaluation-harness/lm_eval/evaluator.py", line 251, in simple_evaluate results = evaluate( File "/home/michael/code/lm-evaluation-harness/lm_eval/utils.py", line 288, in _wrapper return fn(*args, **kwargs) File "/home/michael/code/lm-evaluation-harness/lm_eval/evaluator.py", line 390, in evaluate resps = getattr(lm, reqtype)(cloned_reqs) File "/home/michael/code/lm-evaluation-harness/lm_eval/models/openai_completions.py", line 263, in generate_until list(sameuntil_chunks(re_ord.get_reordered(), self.batch_size)), File "/home/michael/code/lm-evaluation-harness/lm_eval/models/openai_completions.py", line 251, in sameuntil_chunks if len(ret) >= size or x[1] != lastuntil: TypeError: '>=' not supported between instances of 'int' and 'str' ``` * Patch QQP prompt (EleutherAI#1661) * TMMLU+ implementation (EleutherAI#1394) * implementation of TMMLU+ * implemented: TMMLU+ ****TMMLU+ : large-scale Traditional chinese Massive Multitask language Understanding**** - 4 categories - STEM - Social Science - Humanities - Other The TMMLU+ dataset, encompassing over 67 subjects and 20160 tasks, is six times larger and more balanced than its predecessor, TMMLU, and includes benchmark results from both closed-source and 20 open-weight Chinese large language models with 1.8B to 72B parameters. However, Traditional Chinese variants continue to underperform compared to major Simplified Chinese models. ```markdown Total number of tasks in the 'test' sets: 20160 Total number of tasks in the 'validation' sets: 2247 Total number of tasks in the 'train' sets: 335 ``` * Remove print from __init__.py There was my mistake in forgetting to remove the debug print from the code. * update: move TMMLU+ config generation program into default * fix: we should use training set as few shots example * update: README for TMMLU+ * update: a small changes of TMMLU+ README file * pre-commit run thought * Add README for TMMLU+ dataset * run precommit * trigger precommit again * trigger precommit again * isort is fussy * isort is fussy * format, again * oops * oops --------- Co-authored-by: lintang <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Anthropic Chat API (EleutherAI#1594) * claude3 * supply for anthropic claude3 * supply for anthropic claude3 * anthropic config changes * add callback options on anthropic * line passed * claude3 tiny change * help anthropic installation * mention sysprompt / being careful with format in readme --------- Co-authored-by: haileyschoelkopf <[email protected]> * correction bug EleutherAI#1664 (EleutherAI#1670) * correction bug EleutherAI#1664 * add any invalid characters for Windows filenames and Unix-like systems see: https://gist.github.com/doctaphred/d01d05291546186941e1b7ddc02034d3?permalink_comment_id=3958715 * Update lm_eval/__main__.py * Update scripts/zeno_visualize.py * fix format --------- Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Update README.md (EleutherAI#1680) * Add delta weights model loading (EleutherAI#1712) * added delta weights * removed debug * readme update * better error handling * autogptq warn * warn update * peft and delta error, explicitly deleting _model_delta * linter fix * Add `neuralmagic` models for `sparseml` and `deepsparse` (EleutherAI#1674) * Add neuralmagic models for SparseML and DeepSparse * Update to latest and add test * Format * Fix list to List * Format * Add deepsparse/sparseml to automated testing * Update pyproject.toml * Update pyproject.toml * Update README * Fixes for dtype and device * Format * Fix test * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> * Address review comments! --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * fix error when appending eot_token_id for generate_until tasks (EleutherAI#1699) * Adding retries and rate limit to toxicity tasks (EleutherAI#1620) * reference `--tasks list` in README (EleutherAI#1726) EleutherAI#1698 * Add XNLIeu: a dataset for cross-lingual NLI in Basque (EleutherAI#1694) * add xnli_eu tasks * update tasks readme * update readme * Fix Parameter Propagation for Tasks that have `include` (EleutherAI#1749) * Update task.py * Update __init__.py * Support individual scrolls datasets (EleutherAI#1740) * Support individual scrolls datasets * Add qmsum context * Fix formatting * Add filter registry decorator (EleutherAI#1750) * Add register_filter decorator * Add register_filter docs * remove duplicated `num_fewshot: 0` (EleutherAI#1769) * Pile 10k new task (EleutherAI#1758) * Add Pile-10k readme * Add Pile-10k task configuration file * Fix m_arc choices (EleutherAI#1760) * Update utils.py This is a 4-choice task, option_e is null for all but 3 samples * Fix options Adaptive choices * add option e * bump multilingual arc version --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * upload new tasks (EleutherAI#1728) * upload new tasks * add readmes * run linters --------- Co-authored-by: haileyschoelkopf <[email protected]> * vllm lora support (EleutherAI#1756) * vllm lora support * remove print * version check, rename lora kwarg * Add option to set OpenVINO config (EleutherAI#1730) * Add option to set OpenVINO config * Use utils.eval_logger for logging * evaluation tracker implementation (EleutherAI#1766) * evaluation tracker implementation * OVModelForCausalLM test fix * typo fix * moved methods args * multiple args in one flag * loggers moved to dedicated dir * improved filename sanitization * eval tracker args fix (EleutherAI#1777) * limit fix (EleutherAI#1785) * remove echo parameter in OpenAI completions API (EleutherAI#1779) * remove echo parameter in OpenAI completions API * remove context length parameter doc string * Fix README: change`----hf_hub_log_args` to `--hf_hub_log_args` (EleutherAI#1776) fix `----hf_hub_log_args` to `--hf_hub_log_args` * Fix bug in setting until kwarg in openai completions (EleutherAI#1784) * Provide ability for custom sampler for ConfigurableTask (EleutherAI#1616) * Added fewshot sampling seeds to evaluator.simple_evaluate signature Way to control seed of fewshot sampling may help with EleutherAI#1591 * Added ability for custom sampler for ConfigurableTask May be set in config like ``` fewshot_config: sampler: !function utils.MyFewshotSampler ``` * explicitly set fewshot random generator seed for HFLM generate_until_task test * add backward compatibility for three args seed setup * save seeds info to logs/reports * Update `--tasks list` option in interface documentation (EleutherAI#1792) * Fix Caching Tests ; Remove `pretrained=gpt2` default (EleutherAI#1775) * link to the example output on the hub (EleutherAI#1798) * Re-add Hendrycks MATH (no sympy checking, no Minerva hardcoded prompt) variant (EleutherAI#1793) * add Hendrycks MATH (no sympy checking) variant * add readmes for MATH tasks * Logging Updates (Alphabetize table printouts, fix eval tracker bug) (EleutherAI#1774) (EleutherAI#1791) * fix auto-batch size bug for seq2seq models * alphabetize task + group tables ; fix eval tracker bug * fix eval tracker bug * Initial integration of the Unitxt to LM eval harness (EleutherAI#1615) * Initial support for Unitxt datasets in LM Eval Harness See https://github.com/IBM/unitxt The script 'generate_yamls.py' creates LM Eval Harness yaml files corresponding to Unitxt datasets specified in the 'unitxt_datasets' file. The glue code required to register Unitxt metrics is in 'unitxt_wrapper.py'. * Added dataset loading check to generate_yaml Improved error messages. * Speed up generate_yaml Added printouts and improved error message * Added output printout * Simplified integration of unitxt datasets Store all the common yaml configuration in a yaml include shared by all datasets of the same task. * Post code review comments - part 1 1. Made sure include files don't end wth 'yaml' so they won't be marked as tasks 2. Added more datasets and tasks (NER, GEC) 3. Added README * Post code review comments - part 2 1. Added install unitxt install option in pyproject.toml: pip install 'lm_eval[unitxt]' 2. Added a check that unitxt is installed and print a clear error message if not * Commited missing pyproject change * Added documentation on adding datasets * More doc changes * add unitxt extra to readme * run precommit --------- Co-authored-by: haileyschoelkopf <[email protected]> * add task for mmlu evaluation in arc multiple choice format (EleutherAI#1745) * add mmlu arc style evaluation * rename arc_style to continuation --------- Co-authored-by: Jonathan Burdge <[email protected]> Co-authored-by: Jonathan Burdge <[email protected]> * Update flag `--hf_hub_log_args` in interface documentation (EleutherAI#1806) * update interface documentation with flag --hf_hub_logs_arg * update interface documentation with flag --hf_hub_logs_arg 2 * Copal task (EleutherAI#1803) * add copal * change name to copal id for clarity and the task name * remove `copal_id...` to yaml to make it work * checkmark on README * change group name to `copal_id` * Adding tinyBenchmarks datasets (EleutherAI#1545) * Add tinyBenchmarks * Add acknowledgements * Add ordering of outputs for data-parallel * Run pre-commit * Add few_shot specifications * Add tinyBenchmarks post-processing * add conditional import ; fix task names --------- Co-authored-by: haileyschoelkopf <[email protected]> * interface doc update (EleutherAI#1807) * Fix links in README guiding to another branch (EleutherAI#1838) * Fix: support PEFT/LoRA with added tokens (EleutherAI#1828) * resize model embeddings * resize only * tokenizer help * load tokenizer before model * add comment and run precommit lint * Add log message Co-authored-by: Hailey Schoelkopf <[email protected]> --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * fixed incorrect check for task type (replace `~` with `not`) (EleutherAI#1865) * fixed docs typos (EleutherAI#1863) * Update polemo2_out.yaml (EleutherAI#1871) * Unpin vllm in dependencies (EleutherAI#1874) * Fix outdated links to the latest links in `docs` (EleutherAI#1876) * [HFLM]Use Accelerate's API to reduce hard-coded CUDA code (EleutherAI#1880) * Fix `batch_size=auto` for HF Seq2Seq models (EleutherAI#1765) (EleutherAI#1790) * fix auto-batch size bug for seq2seq models * run linter * Fix Brier Score (EleutherAI#1847) `gold_one_hot` needs to follow the dimension of predictions so that it still works when `--limit` is used and the indexes in gold does not cover all gold indexes. * Fix for bootstrap_iters = 0 case (EleutherAI#1715) (EleutherAI#1789) * add handling for bootstrap_iters=0 case * add more detail to docstring * run precommit * add mmlu tasks from pile-t5 (EleutherAI#1710) * add mmlu tasks from pile-t5 * Update _mmlu_flan_cot_fewshot_template_yaml * Update _mmlu_flan_cot_zeroshot_template_yaml * Update _mmlu_flan_generative_template_yaml * Update _mmlu_flan_loglikelihood_template_yaml * Update _default_template_yaml --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Bigbench fix (EleutherAI#1686) * edit process multiple-choice * split template yaml * remove * modified multiple_choice tasks * udpate * Update multiple_choice_template_b_yaml * Update multiple_choice_template_a_yaml --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Rename `lm_eval.logging -> lm_eval.loggers` (EleutherAI#1858) * rename lm_eval.logging module * fix evaluation tracker args * Updated vllm imports in vllm_causallms.py (EleutherAI#1890) * Reorder vllm imports in vllm_causallms.py * Update vllm_causallms.py * [HFLM]Add support for Ascend NPU (EleutherAI#1886) * [HFLM]Add support for Ascend NPU Co-authored-by: jiaqiw09 <[email protected]> Co-authored-by: zhabuye <[email protected]> * bump accelerate dependency version to 0.26.0 for NPU compat. --------- Co-authored-by: jiaqiw09 <[email protected]> Co-authored-by: zhabuye <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> * `higher_is_better` tickers in output table (EleutherAI#1893) * Higher is better tickers in output table * add extra check for `higher_is_better` not being None already * Update lm_eval/evaluator.py * fixup format I messed up * add comment (and retrigger tests) --------- Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Add dataset card when pushing to HF hub (EleutherAI#1898) * dataset card initial * few fixes * adds groups for math, mmlu, gpqa * added summary agrs * moved sanitize_list to utils * readme update * recreate metadata moved * multiple model support * results latest split fix * readme update and small refactor * fix grouping * add comments * added pathlib * corrected pathlib approach * check whether to create a metadata card * convert posix paths to str * default hf org from token * hf token value error * Add logs after successful upload * logging updates * dataset card example in the readme --------- Co-authored-by: Nathan Habib <[email protected]> Co-authored-by: Alina Lozovskaia <[email protected]> * Making hardcoded few shots compatible with the chat template mechanism (EleutherAI#1895) * init test 1 * fix * this format seems to be working - need to update all other tasks with the new format * bbh with few shot format * fix fewshot bbh * add mmlu flan cot * samples of cot * kmmlu * fix gsm8k * update keys for mmlu * minerva math * bbh * fix * fix samples * small fixes to templates * last prompt format change * fixing prompt * fixed minerva math format * rm accidental commited file * added doc for few shot samples * Update lm_eval/loggers/evaluation_tracker.py * Update lm_eval/loggers/evaluation_tracker.py * Update docs/new_task_guide.md Co-authored-by: Hailey Schoelkopf <[email protected]> * added check in sampler per code review * added the system from a function, plus an example in minerva math * style * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> * fix unit tests 1 * forcing use of test split --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Try to make existing tests run little bit faster (EleutherAI#1905) * Fix fewshot seed only set when overriding num_fewshot (EleutherAI#1914) Fix EleutherAI#1906 * Complete task list from pr 1727 (EleutherAI#1901) * added tasks and task family descriptors * continue work on task list w/ links; slightly reorganize README * Apply suggestions from code review * Rename file so that it'll preview in Github when viewing lm_eval/tasks folder * Update new_task_guide.md * Update README.md * run linter * Add language column to task table; Add missing tasks to task table; fix nq_open and storycloze READMEs * fix typo * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> * apply format --------- Co-authored-by: Harish Vadaparty <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Add chat template (EleutherAI#1873) * initial chat template * tokenizer attribute check * variable rename * interface update * system instruction * system inst default update * fewshot as multiturn * typing update * indent update * added comments * Adding a fewshot in a more readable way * linting * Moved apply chat template to LM * multiturn alternation fix * cache key update * apply chat template method fix * add system prompt hash to cache_key * tokenizer name property for cache_key * property name fix * linting backward compatibility fix * docs and errors update * add documentation on adding chat template compatibility to model_guide * fewshot as multiturn check fix * saving system inst and chat template in results * eval tracker update * docs update * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> --------- Co-authored-by: haileyschoelkopf <[email protected]> Co-authored-by: Clémentine Fourrier <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> * Multiple Choice Questions and Large Languages Models: A Case Study with Fictional Medical Data (EleutherAI#1867) * glianorex tasks * Create README.md * Update README.md * Update README.md * fix formatting * fix internal formatting * Modify pre-commit hook to check merge conflicts accidentally committed not at current merge commit (EleutherAI#1927) * [add] fld logical formula task (EleutherAI#1931) * Add new Lambada translations (EleutherAI#1897) * added tasks and task family descriptors * configs for the new lambada translations * continue work on task list w/ links; slightly reorganize README * Apply suggestions from code review * Rename file so that it'll preview in Github when viewing lm_eval/tasks folder * Update new_task_guide.md * Update README.md * run linter * Add language column to task table; Add missing tasks to task table; fix nq_open and storycloze READMEs * fix typo * update `lm_eval/tasks/README.md` with task description --------- Co-authored-by: Harish Vadaparty <[email protected]> Co-authored-by: anthony <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Implement NoticIA (EleutherAI#1912) * Noticia * test * Final testes implementation * Fixes * Fix linters * Add The Arabic version of the PICA benchmark (EleutherAI#1917) * Update siqa.yaml (EleutherAI#1909) * Update basque-glue (EleutherAI#1913) * Update README.md * Update bec.yaml * Update bhtc.yaml * Update coref.yaml * Update qnli.yaml * Update vaxx.yaml * Update wic.yaml * Test output table layout consistency (EleutherAI#1916) * sort metrics in output table * update docstring in `consolidate_results` * add tests for verifying consistency of table output * update tests to account for floating point inconsistencies * updated tests based on `pythia-14m` * Update __main__.py (EleutherAI#1939) * Add the Arabic version with refactor to Arabic pica to be in alghafa folder (EleutherAI#1940) * Results filenames handling fix (EleutherAI#1926) * results filenames handling moved to utils * zeno results handling fix * tasks_for_model backward compatibility * results files logic moved to tasks_for_model * moved sanitize_model_name to utils * Remove AMMLU Due to Translation (EleutherAI#1948) * Update README.md * Delete lm_eval/tasks/ammlu directory * add include_defaults kwarg to taskmanager, add tests for include_path (EleutherAI#1856) * add hacky add_bos_token forcing for Gemma to VLLM too (EleutherAI#1857) * Update interface.md (EleutherAI#1955) * Fix self.max_tokens in anthropic_llms.py (EleutherAI#1848) Fix bug where `self.max_tokens` was not set * `samples` is newline delimited (EleutherAI#1930) * `samples` is newline delimited * updated git and pre-commit * appease pre-commit * nit * Revert back for now * Revert for now --------- Co-authored-by: Lintang Sutawika <[email protected]> * Fix `--gen_kwargs` and VLLM (`temperature` not respected) (EleutherAI#1800) * Update vllm_causallms.py * adjust --------- Co-authored-by: lintangsutawika <[email protected]> * make write_out.py explicitly error if no splits match (EleutherAI#1796) Co-authored-by: lintangsutawika <[email protected]> * fix: add directory filter to os.walk to ignore 'ipynb_checkpoints' (EleutherAI#1956) * fix: add filter to os.walk to ignore 'ipynb_checkpoints * Update __init__.py * Update __init__.py --------- Co-authored-by: Lintang Sutawika <[email protected]> * add trust_remote_code for piqa (EleutherAI#1983) Signed-off-by: changwangss <[email protected]> * Fix self assignment in neuron_optimum.py (EleutherAI#1990) * [New Task] Add Paloma benchmark (EleutherAI#1928) * init paloma benchmark * pre-process in utils function * add `task_alias` * updated task aliases * Update paloma_dolma-v1_5.yaml * Update paloma_twitterAAE_HELM_fixed.yaml * Update paloma_dolma_100_programing_languages.yaml --------- Co-authored-by: Lintang Sutawika <[email protected]> * Fix Paloma Template yaml (EleutherAI#1993) * init paloma benchmark * pre-process in utils function * add `task_alias` * updated task aliases * Update paloma_dolma-v1_5.yaml * Update paloma_twitterAAE_HELM_fixed.yaml * Update paloma_dolma_100_programing_languages.yaml * update on names * fix paloma template issue --------- Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: Lintang Sutawika <[email protected]> * Log `fewshot_as_multiturn` in results files (EleutherAI#1995) * log fewshot_as_multiturn in general tracker args * Update evaluator.py --------- Co-authored-by: Lintang Sutawika <[email protected]> * Added ArabicMMLU (EleutherAI#1987) * Added ArabicMMLU * Rename `ammlu` to `arabicmmlu` * Fix Datasets `--trust_remote_code` (EleutherAI#1998) * Add BertaQA dataset tasks (EleutherAI#1964) * add bertaqa tasks * rename basquetrivia-->bertaqa ; make template stub not .yaml * add bertaqa entry to lm_eval/tasks/README.md --------- Co-authored-by: haileyschoelkopf <[email protected]> * Fix precommit hook, update run_models.sh * Rename main mmlu ru config * Add ru continuation version --------- Signed-off-by: changwangss <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: khalil <[email protected]> Co-authored-by: Lintang Sutawika <[email protected]> Co-authored-by: Alex Bäuerle <[email protected]> Co-authored-by: Wongboo <[email protected]> Co-authored-by: achervyakov <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> Co-authored-by: Eitan Turok <[email protected]> Co-authored-by: Rylan Schaeffer <[email protected]> Co-authored-by: Vicki Boykis <[email protected]> Co-authored-by: Lintang Sutawika <[email protected]> Co-authored-by: kwrobel.eth <[email protected]> Co-authored-by: Nouf M. Alotaibi <[email protected]> Co-authored-by: Haonan Li <[email protected]> Co-authored-by: Baber Abbasi <[email protected]> Co-authored-by: WoosungMyung <[email protected]> Co-authored-by: Sergio Perez <[email protected]> Co-authored-by: Or Sharir <[email protected]> Co-authored-by: Julen Etxaniz <[email protected]> Co-authored-by: Michael Goin <[email protected]> Co-authored-by: ZoneTwelve <[email protected]> Co-authored-by: Seungwoo Ryu <[email protected]> Co-authored-by: nicho2 <[email protected]> Co-authored-by: KonradSzafer <[email protected]> Co-authored-by: Sergio Perez <[email protected]> Co-authored-by: sator-labs <[email protected]> Co-authored-by: Brian Vaughan <[email protected]> Co-authored-by: giorgossideris <[email protected]> Co-authored-by: Nikita Lozhnikov <[email protected]> Co-authored-by: Chujie Zheng <[email protected]> Co-authored-by: Gabriel Mukobi <[email protected]> Co-authored-by: Zehan Li <[email protected]> Co-authored-by: Simran Arora <[email protected]> Co-authored-by: bcicc <[email protected]> Co-authored-by: Helena Kloosterman <[email protected]> Co-authored-by: Muhammad Bin Usman <[email protected]> Co-authored-by: ciaranby <[email protected]> Co-authored-by: LSinev <[email protected]> Co-authored-by: aditya thomas <[email protected]> Co-authored-by: Yoav Katz <[email protected]> Co-authored-by: jonabur <[email protected]> Co-authored-by: Jonathan Burdge <[email protected]> Co-authored-by: Jonathan Burdge <[email protected]> Co-authored-by: Edd <[email protected]> Co-authored-by: Lucas Weber <[email protected]> Co-authored-by: Nick Doiron <[email protected]> Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: zhabuye <[email protected]> Co-authored-by: Edward Gan <[email protected]> Co-authored-by: DongGeon Lee <[email protected]> Co-authored-by: Huazhong Ji <[email protected]> Co-authored-by: jiaqiw09 <[email protected]> Co-authored-by: zhabuye <[email protected]> Co-authored-by: Nathan Habib <[email protected]> Co-authored-by: Alina Lozovskaia <[email protected]> Co-authored-by: Clémentine Fourrier <[email protected]> Co-authored-by: anthony-dipofi <[email protected]> Co-authored-by: Harish Vadaparty <[email protected]> Co-authored-by: Maxime <[email protected]> Co-authored-by: MorishT <[email protected]> Co-authored-by: Iker García-Ferrero <[email protected]> Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: Sadra Barikbin <[email protected]> Co-authored-by: johnwee1 <[email protected]> Co-authored-by: Wang, Chang <[email protected]> Co-authored-by: Yazeed Alnumay <[email protected]>
mariagrandury
pushed a commit
to somosnlp/lm-evaluation-harness
that referenced
this pull request
Jul 25, 2024
…#1800) * Update vllm_causallms.py * adjust --------- Co-authored-by: lintangsutawika <[email protected]>
mansicer
added a commit
to mansicer/lm-evaluation-harness
that referenced
this pull request
Aug 1, 2024
* Fix: support PEFT/LoRA with added tokens (EleutherAI#1828) * resize model embeddings * resize only * tokenizer help * load tokenizer before model * add comment and run precommit lint * Add log message Co-authored-by: Hailey Schoelkopf <[email protected]> --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * fixed incorrect check for task type (replace `~` with `not`) (EleutherAI#1865) * fixed docs typos (EleutherAI#1863) * Update polemo2_out.yaml (EleutherAI#1871) * Unpin vllm in dependencies (EleutherAI#1874) * Fix outdated links to the latest links in `docs` (EleutherAI#1876) * [HFLM]Use Accelerate's API to reduce hard-coded CUDA code (EleutherAI#1880) * Fix `batch_size=auto` for HF Seq2Seq models (EleutherAI#1765) (EleutherAI#1790) * fix auto-batch size bug for seq2seq models * run linter * Fix Brier Score (EleutherAI#1847) `gold_one_hot` needs to follow the dimension of predictions so that it still works when `--limit` is used and the indexes in gold does not cover all gold indexes. * Fix for bootstrap_iters = 0 case (EleutherAI#1715) (EleutherAI#1789) * add handling for bootstrap_iters=0 case * add more detail to docstring * run precommit * add mmlu tasks from pile-t5 (EleutherAI#1710) * add mmlu tasks from pile-t5 * Update _mmlu_flan_cot_fewshot_template_yaml * Update _mmlu_flan_cot_zeroshot_template_yaml * Update _mmlu_flan_generative_template_yaml * Update _mmlu_flan_loglikelihood_template_yaml * Update _default_template_yaml --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Bigbench fix (EleutherAI#1686) * edit process multiple-choice * split template yaml * remove * modified multiple_choice tasks * udpate * Update multiple_choice_template_b_yaml * Update multiple_choice_template_a_yaml --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Rename `lm_eval.logging -> lm_eval.loggers` (EleutherAI#1858) * rename lm_eval.logging module * fix evaluation tracker args * Updated vllm imports in vllm_causallms.py (EleutherAI#1890) * Reorder vllm imports in vllm_causallms.py * Update vllm_causallms.py * [HFLM]Add support for Ascend NPU (EleutherAI#1886) * [HFLM]Add support for Ascend NPU Co-authored-by: jiaqiw09 <[email protected]> Co-authored-by: zhabuye <[email protected]> * bump accelerate dependency version to 0.26.0 for NPU compat. --------- Co-authored-by: jiaqiw09 <[email protected]> Co-authored-by: zhabuye <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> * `higher_is_better` tickers in output table (EleutherAI#1893) * Higher is better tickers in output table * add extra check for `higher_is_better` not being None already * Update lm_eval/evaluator.py * fixup format I messed up * add comment (and retrigger tests) --------- Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Add dataset card when pushing to HF hub (EleutherAI#1898) * dataset card initial * few fixes * adds groups for math, mmlu, gpqa * added summary agrs * moved sanitize_list to utils * readme update * recreate metadata moved * multiple model support * results latest split fix * readme update and small refactor * fix grouping * add comments * added pathlib * corrected pathlib approach * check whether to create a metadata card * convert posix paths to str * default hf org from token * hf token value error * Add logs after successful upload * logging updates * dataset card example in the readme --------- Co-authored-by: Nathan Habib <[email protected]> Co-authored-by: Alina Lozovskaia <[email protected]> * Making hardcoded few shots compatible with the chat template mechanism (EleutherAI#1895) * init test 1 * fix * this format seems to be working - need to update all other tasks with the new format * bbh with few shot format * fix fewshot bbh * add mmlu flan cot * samples of cot * kmmlu * fix gsm8k * update keys for mmlu * minerva math * bbh * fix * fix samples * small fixes to templates * last prompt format change * fixing prompt * fixed minerva math format * rm accidental commited file * added doc for few shot samples * Update lm_eval/loggers/evaluation_tracker.py * Update lm_eval/loggers/evaluation_tracker.py * Update docs/new_task_guide.md Co-authored-by: Hailey Schoelkopf <[email protected]> * added check in sampler per code review * added the system from a function, plus an example in minerva math * style * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> * fix unit tests 1 * forcing use of test split --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Try to make existing tests run little bit faster (EleutherAI#1905) * Fix fewshot seed only set when overriding num_fewshot (EleutherAI#1914) Fix EleutherAI#1906 * Complete task list from pr 1727 (EleutherAI#1901) * added tasks and task family descriptors * continue work on task list w/ links; slightly reorganize README * Apply suggestions from code review * Rename file so that it'll preview in Github when viewing lm_eval/tasks folder * Update new_task_guide.md * Update README.md * run linter * Add language column to task table; Add missing tasks to task table; fix nq_open and storycloze READMEs * fix typo * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> * apply format --------- Co-authored-by: Harish Vadaparty <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Add chat template (EleutherAI#1873) * initial chat template * tokenizer attribute check * variable rename * interface update * system instruction * system inst default update * fewshot as multiturn * typing update * indent update * added comments * Adding a fewshot in a more readable way * linting * Moved apply chat template to LM * multiturn alternation fix * cache key update * apply chat template method fix * add system prompt hash to cache_key * tokenizer name property for cache_key * property name fix * linting backward compatibility fix * docs and errors update * add documentation on adding chat template compatibility to model_guide * fewshot as multiturn check fix * saving system inst and chat template in results * eval tracker update * docs update * Apply suggestions from code review Co-authored-by: Hailey Schoelkopf <[email protected]> --------- Co-authored-by: haileyschoelkopf <[email protected]> Co-authored-by: Clémentine Fourrier <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> * Multiple Choice Questions and Large Languages Models: A Case Study with Fictional Medical Data (EleutherAI#1867) * glianorex tasks * Create README.md * Update README.md * Update README.md * fix formatting * fix internal formatting * Modify pre-commit hook to check merge conflicts accidentally committed not at current merge commit (EleutherAI#1927) * [add] fld logical formula task (EleutherAI#1931) * Add new Lambada translations (EleutherAI#1897) * added tasks and task family descriptors * configs for the new lambada translations * continue work on task list w/ links; slightly reorganize README * Apply suggestions from code review * Rename file so that it'll preview in Github when viewing lm_eval/tasks folder * Update new_task_guide.md * Update README.md * run linter * Add language column to task table; Add missing tasks to task table; fix nq_open and storycloze READMEs * fix typo * update `lm_eval/tasks/README.md` with task description --------- Co-authored-by: Harish Vadaparty <[email protected]> Co-authored-by: anthony <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Implement NoticIA (EleutherAI#1912) * Noticia * test * Final testes implementation * Fixes * Fix linters * Add The Arabic version of the PICA benchmark (EleutherAI#1917) * Update siqa.yaml (EleutherAI#1909) * Update basque-glue (EleutherAI#1913) * Update README.md * Update bec.yaml * Update bhtc.yaml * Update coref.yaml * Update qnli.yaml * Update vaxx.yaml * Update wic.yaml * Test output table layout consistency (EleutherAI#1916) * sort metrics in output table * update docstring in `consolidate_results` * add tests for verifying consistency of table output * update tests to account for floating point inconsistencies * updated tests based on `pythia-14m` * Update __main__.py (EleutherAI#1939) * Add the Arabic version with refactor to Arabic pica to be in alghafa folder (EleutherAI#1940) * Results filenames handling fix (EleutherAI#1926) * results filenames handling moved to utils * zeno results handling fix * tasks_for_model backward compatibility * results files logic moved to tasks_for_model * moved sanitize_model_name to utils * Remove AMMLU Due to Translation (EleutherAI#1948) * Update README.md * Delete lm_eval/tasks/ammlu directory * add include_defaults kwarg to taskmanager, add tests for include_path (EleutherAI#1856) * add hacky add_bos_token forcing for Gemma to VLLM too (EleutherAI#1857) * Update interface.md (EleutherAI#1955) * Fix self.max_tokens in anthropic_llms.py (EleutherAI#1848) Fix bug where `self.max_tokens` was not set * `samples` is newline delimited (EleutherAI#1930) * `samples` is newline delimited * updated git and pre-commit * appease pre-commit * nit * Revert back for now * Revert for now --------- Co-authored-by: Lintang Sutawika <[email protected]> * Fix `--gen_kwargs` and VLLM (`temperature` not respected) (EleutherAI#1800) * Update vllm_causallms.py * adjust --------- Co-authored-by: lintangsutawika <[email protected]> * make write_out.py explicitly error if no splits match (EleutherAI#1796) Co-authored-by: lintangsutawika <[email protected]> * fix: add directory filter to os.walk to ignore 'ipynb_checkpoints' (EleutherAI#1956) * fix: add filter to os.walk to ignore 'ipynb_checkpoints * Update __init__.py * Update __init__.py --------- Co-authored-by: Lintang Sutawika <[email protected]> * add trust_remote_code for piqa (EleutherAI#1983) Signed-off-by: changwangss <[email protected]> * Fix self assignment in neuron_optimum.py (EleutherAI#1990) * [New Task] Add Paloma benchmark (EleutherAI#1928) * init paloma benchmark * pre-process in utils function * add `task_alias` * updated task aliases * Update paloma_dolma-v1_5.yaml * Update paloma_twitterAAE_HELM_fixed.yaml * Update paloma_dolma_100_programing_languages.yaml --------- Co-authored-by: Lintang Sutawika <[email protected]> * Fix Paloma Template yaml (EleutherAI#1993) * init paloma benchmark * pre-process in utils function * add `task_alias` * updated task aliases * Update paloma_dolma-v1_5.yaml * Update paloma_twitterAAE_HELM_fixed.yaml * Update paloma_dolma_100_programing_languages.yaml * update on names * fix paloma template issue --------- Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: Lintang Sutawika <[email protected]> * Log `fewshot_as_multiturn` in results files (EleutherAI#1995) * log fewshot_as_multiturn in general tracker args * Update evaluator.py --------- Co-authored-by: Lintang Sutawika <[email protected]> * Added ArabicMMLU (EleutherAI#1987) * Added ArabicMMLU * Rename `ammlu` to `arabicmmlu` * Fix Datasets `--trust_remote_code` (EleutherAI#1998) * Add BertaQA dataset tasks (EleutherAI#1964) * add bertaqa tasks * rename basquetrivia-->bertaqa ; make template stub not .yaml * add bertaqa entry to lm_eval/tasks/README.md --------- Co-authored-by: haileyschoelkopf <[email protected]> * add tokenizer logs info (EleutherAI#1731) * add tokenizer logs info * add no tokenizer case * Update lm_eval/logging_utils.py Co-authored-by: Hailey Schoelkopf <[email protected]> * Update lm_eval/logging_utils.py Co-authored-by: Hailey Schoelkopf <[email protected]> * add updates * fix conflict --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Hotfix breaking import (EleutherAI#2015) * add arc_challenge_mt (EleutherAI#1900) * add arc_challenge_mt * add README * add icelandic * Remove `LM` dependency from `build_all_requests` (EleutherAI#2011) * refactored `lm.apply_chat_template` * nit * fix weird type error * fixed! * skip failing test * pre-commit run all * add type hints * nit * nit * fixup * Added CommonsenseQA task (EleutherAI#1721) * Initial configuration * Using the validation set for the test set, because the test set on HF doesn't have labels * Probably just makes more sense to have validation be validation * fix format ; add docs to tasks/README.md * fix format --------- Co-authored-by: haileyschoelkopf <[email protected]> * Factor out LM-specific tests (EleutherAI#1859) * separate out optimum/neuralmagic tests to separate job * fix vllm tests * fix bug in --trust_remote_code * use datasets.config instead intentionally * fix remote code issue? * Update interface.md (EleutherAI#1982) * Update interface.md update interface to remove link to really outdated commit of evaluator.py * switch to relative referencing? * Update interface.md --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * Fix `trust_remote_code`-related test failures (EleutherAI#2024) * make MMLU trust remote code to fix tests * remove trust remote code * Fixes scrolls task bug with few_shot examples (EleutherAI#2003) Bug: ``` python -m scripts.write_out --task scrolls_quality --output_base_path ~/workspace/ Traceback (most recent call last): File "<frozen runpy>", line 198, in _run_module_as_main File "<frozen runpy>", line 88, in _run_code File "/lm-evaluation-harness/scripts/write_out.py", line 92, in <module> main() File "/lm-evaluation-harness/scripts/write_out.py", line 51, in main task_dict = tasks.get_task_dict(task_names, task_manager) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/lm-evaluation-harness/lm_eval/tasks/__init__.py", line 423, in get_task_dict task_name_from_string_dict = task_manager.load_task_or_group( ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/lm-evaluation-harness/lm_eval/tasks/__init__.py", line 271, in load_task_or_group collections.ChainMap(*map(self._load_individual_task_or_group, task_list)) File "/lm-evaluation-harness/lm_eval/tasks/__init__.py", line 162, in _load_individual_task_or_group return load_task(task_config, task=name_or_config, group=parent_name) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/lm-evaluation-harness/lm_eval/tasks/__init__.py", line 148, in load_task task_object = config["class"]() ^^^^^^^^^^^^^^^^^ File "/lm-evaluation-harness/lm_eval/tasks/scrolls/task.py", line 120, in __init__ super().__init__() File "/lm-evaluation-harness/lm_eval/api/task.py", line 703, in __init__ self._config = TaskConfig(**config) ^^^^^^^^^^^^^^^^^^^^ TypeError: lm_eval.api.task.TaskConfig() argument after ** must be a mapping, not NoneType ``` * fix cache (EleutherAI#2037) * Add chat template to `vllm` (EleutherAI#2034) * add chat template * refactor token padding * nit * nit * check on failing test * check transformers version * remove transformers pin * add ids to test * nit * fixup * fix bos bug * nit * fixup! fix bos bug * increase tolerance for table test * don't detokenize vllm logprobs * Update lm_eval/models/utils.py Co-authored-by: Hailey Schoelkopf <[email protected]> * pre-commit run --all-files --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * fail gracefully upon tokenizer logging failure (EleutherAI#2038) * ship with exact_match function already used ; don't call evaluate.load() on import (EleutherAI#2045) * update to v0.4.3 (EleutherAI#2046) * fix wandb logger module import in example (EleutherAI#2041) * Fix strip whitespace filter (EleutherAI#2048) * batch commit * :Revert "batch commit" This reverts commit d859d1c. * batch commit * checkout from main * checkout from main * checkout from main * checkout from main * checkout from main * cleanup * cleanup * cleanup * cleanup * cleanup * cleanup * update gemma-2 default BOS behavior (EleutherAI#2049) * Update hellaswag.yaml (EleutherAI#2029) * Adds Open LLM Leaderboard Taks (EleutherAI#2047) * adds leaderboard tasks * Delete lm_eval/tasks/leaderboard/leaderboard_chat_template.yaml * add readme * Delete lm_eval/tasks/leaderboard/mmlu_pro/mmlu_pro_chat_template.yaml * modify readme * fix bbh task * fix bbh salient task * modify the readme * Delete lm_eval/tasks/leaderboard/ifeval/README.md * Delete lm_eval/tasks/leaderboard/math/README.md * add leaderboard to the tasks repertory * add anouncment about new leaderbaord tasks * linting * Update README.md Co-authored-by: Hailey Schoelkopf <[email protected]> * installs ifeval dependency in new_task github workflow --------- Co-authored-by: Nathan Habib <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> * EleutherAI#1442 inverse scaling tasks implementation (EleutherAI#1589) * initial_implementation (test has to be proceeded) * minor fix * revised task name and implemented new task * minor fixes * new tasks implement * minor fix * added 'prompt injection' task * delete prompt injection task (will be implemented at next PR) * trust remote code * Update lm_eval/tasks/inverse_scaling/README.md Co-authored-by: Hailey Schoelkopf <[email protected]> * added readme * Update lm_eval/tasks/README.md * Update lm_eval/tasks/inverse_scaling/_inverse_scaling_mc_yaml * Update lm_eval/tasks/inverse_scaling/README.md Co-authored-by: Hailey Schoelkopf <[email protected]> * Update lm_eval/tasks/inverse_scaling/_inverse_scaling_mc_yaml Co-authored-by: Hailey Schoelkopf <[email protected]> * Update README.md * precommit? * run precommit on readme --------- Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Fix TypeError in samplers.py by converting int to str (EleutherAI#2074) Co-authored-by: yhjo <[email protected]> * Group agg rework (EleutherAI#1741) * add greoup_config arg * add a group config that allows disabling table for group score and group aggregate in general * fixed size configuration * adjust config * add group config * adjust mmlu to use group_config * fixed args input in aggregate_subtask_metrics * fixed issues related to printing alias of group and updated yaml * update all mmlu variants to include group_config * edit format * modify mmlu tasks * adjust group to also be a configurable group * add configurable group * simplify get_task_list * adjust group scoring with using ConfigurableGroup * adjust args * update mmlu * update mmlu * update to work with new group and task configuration * readd group_agg * readd files * move prepare_print_tasks to evaluator_utils * sort set to False by default, fix predict_only arg * add version for groups * reversed task list * update additional condition when loading a group in a group yaml * update truthfulqa * add description regarding tags replacing group * replace group to tag * fixed conditional statement * remove warning * update loading of task group and newly added tags * reformat with pre-commit * fixed info log * update * fix bug * fix bug * use task id to differentiate tasks * convert all groups to configurable groups * use task_id * reformat * add task_id for python tasks as well * add task_id for python tasks as well * add task_id for python tasks as well * revert truthfulqa * revert mmlu tasks * new mmlu config * new group config parameter `tag_to_task` * Update truthfulqa_mc2.yaml * reformate * add _process_group_config * adjust task_id * add get_subtask_list function to get proper subtask list * group config to_dict update * remove tag check * update mmlu * fix config passing issues * add test yaml * format fix * add documentation * corner case for single tag being called * fix indentation * formatting * update all mmlu variants * Update docs/task_guide.md Co-authored-by: Hailey Schoelkopf <[email protected]> * remove group_alias * Update docs/task_guide.md Co-authored-by: Hailey Schoelkopf <[email protected]> * remove version for metadata * Update docs/task_guide.md Co-authored-by: Hailey Schoelkopf <[email protected]> * update mmlu/ * removed " " in make_table * change how aggregate_metric is loaded * change how aggregate_metric is loaded * update aggregate_metric arg * update format * update format * some docs fixes * add groups for agieval, aexams, aclue * add more explicit aggregation groups * add more groupings / tags distinctions * add more groupings * more groupings * add many explicit group configs * add many explicit group configs * add more explicit group configs * add more explicit group configs * add more error msgs, agg_metric -> agg_metric_list * some docs updates * update task_id to be updateable and uses group:task format * make KMMLU a tag for now * update docs * don't duplicate task names * fix merge conflicts? * giving this a try * clean up diff * switch mmlu variants over to using * don't use to-be-deprecated group: config field in overview notebook * Python tasks which subclass ConfigurableTask now run * update mmlu * pre-commit format * fixed sorting for multi-level printing * move group api to separate file * fix bbh aggregation filter usage * track api/group.py * adjust group and tags loading * make explicit group configs for leaderboard and other newer tasks * fix arabicmmlu * update * change arabicmmlu template name??? * update group alias * fix printing bugs * check table printing is correct ; update tests * use mmlu_stem to have a group included in print tests --------- Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * we run with bootstrap_iters=0 for printing tests (EleutherAI#2080) * Easier unitxt tasks loading and removal of unitxt library dependancy (EleutherAI#1933) * Updated unitxt loading Signed-off-by: Elron Bandel <[email protected]> * Revert change to general Readme Signed-off-by: Elron Bandel <[email protected]> * Adjust fda,squadv2,squad_completion and swde to work accept config in the constructor Signed-off-by: Elron Bandel <[email protected]> * Fix scrolls Signed-off-by: elronbandel <[email protected]> * Update documentation Signed-off-by: elronbandel <[email protected]> * Enforce backward compatability Signed-off-by: elronbandel <[email protected]> * Format unitxt class Signed-off-by: elronbandel <[email protected]> --------- Signed-off-by: Elron Bandel <[email protected]> Signed-off-by: elronbandel <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> * Allow gating EvaluationTracker HF Hub results; customizability (EleutherAI#2051) * batch commit * :Revert "batch commit" This reverts commit d859d1c. * batch commit * checkout from main * checkout from main * checkout from main * checkout from main * checkout from main * cleanup * cleanup * cleanup * cleanup * cleanup * cleanup eval results * cleanup * add check for gated repo * fix jsonline issue * fix * add try catch when gating the details repo * add doc * adds back hub_repo_name * readds hub repo name * Minor doc fix: leaderboard README.md missing mmlu-pro group and task (EleutherAI#2075) leaderboard README.md missing mmlu-pro group and task * fix: utf-8 encoding for logged sample files was missing (EleutherAI#2082) * Update utils.py (EleutherAI#2085) Group Configs with no aggregation will print a empty space as the score for result table. Example ``` | Tasks |Version|Filter|n-shot| Metric | |Value | |Stderr| |--------------|-------|------|-----:|--------|---|-----:|---|-----:| |group | N/A| | | | | | | | | - task 0 |Yaml |none | 0|acc |↑ |0.4000|± |0.0910| | - task 1 |Yaml |none | 0|acc |↑ |0.3333|± |0.0875| | - task 2 |Yaml |none | 0|acc |↑ |0.2667|± |0.0821| | - task 3 |Yaml |none | 0|acc |↑ |0.3333|± |0.0875| ``` So the `v` variable in the `make_table` needs to check if the value is a float or a string. * batch_size may be str if 'auto' is specified (EleutherAI#2084) * Prettify lm_eval --tasks list (EleutherAI#1929) * add and ; move task list newline logic to new TaskManager.list_all_tasks() method * format table list into markdown table; add config location column * add Output Type column * add logic for printing table of tags separately * merge with main and fix conflicts ; update docstrings --------- Co-authored-by: haileyschoelkopf <[email protected]> * make RougeScorer only initialized once (EleutherAI#2090) * Update default.yaml (EleutherAI#2092) * Add new dataset MMLU-SR tasks (EleutherAI#2032) * add mmlusr tasks * renamed all tasks names in mmlusr * edit format and readme * added mmlu_sr * mmlu_sr -> mmlusr * update --------- Co-authored-by: lintangsutawika <[email protected]> * Irokobench: Benchmark Dataset for African languages (EleutherAI#2042) * add afrixnli to task * add chat completion * remove chat completion -untested * afrimmlu added * afrimmlu folder update * afrimmlu folder update * updated prompt * remove print * add afrimgsm -direct * add squad metric * fix bash script * remove direct util, update common yaml * remove print * add few show. metric fixes * fix direct path, add bash script for gpt models * added transate test * update afrixnli tasks * update afrixnli tasks * update metrics for afrixnli * prompt translations fix * prompt translations fix * filter and metric fix -mgsm * remove squad metric * remove squad metric * add f1 score to mgsm * add f1 score to mgsm * update native-direct with lin * change f1 function * add lin to utils * add utils * remove test limit * remove test configs * add swahili to mmlu * change eng to ewe in ewe yaml mmlu * add squad metric to mgsm, remove whitespace filter * added translate test * added afrixnli_translate * fix exact match valueError * fix exact match valueError * restructure mmlu folder * spacing * remove afrimmlu_translate folder * add utility * format task name, clean ups * modefied mgsm * update on afrimgsm * update on afrimgsm * removed utils * other mgsm varieties * other mgsm varieties * adding trasnslate direct * Update translate_direct_yaml * add manual xnli prompt, add multichoice for openai models, and adapt multichoice metric for openai model * edit for open models * Update translate_direct_yaml * add verbalizer for xnli * change xnli from multiple choice to generate * add manual accuracy scores * revert xnli to multiple choice * change afrimgsm utils * revert xnli to multiple_choice * cleanups and readmes * remove openai fixes and unused regex * pr review changes * revert metrics.py, task.py and extraction.py to main version --------- Co-authored-by: Israel Abebe Azime <[email protected]> Co-authored-by: Israel Abebe Azime <[email protected]> * docs: remove trailing sentence from contribution doc (EleutherAI#2098) Signed-off-by: Nathan Weinberg <[email protected]> * Added MedConceptsQA Benchmark (EleutherAI#2010) * Added MedConceptsQA Benchmark * pre-commit factor * update group name * update in naming * changed name * Changed mcqa to med_concepts_qa prefix * Added med_concepts_qa to README.md * Changed config files according the new format * Updated README --------- Co-authored-by: lintangsutawika <[email protected]> * make recurrent_gemma model types included in the force-BOS case (EleutherAI#2105) * formatting (EleutherAI#2104) * docs: align local test command to match CI (EleutherAI#2100) Also add 'test_logs/' to .gitignore Signed-off-by: Nathan Weinberg <[email protected]> * Fixed colon in Belebele _default_template_yaml (EleutherAI#2111) * [python] fix haerae tasks (EleutherAI#2112) * fix: broken discord link in CONTRIBUTING.md (EleutherAI#2114) Signed-off-by: Nathan Weinberg <[email protected]> * docs: update truthfulqa tasks (EleutherAI#2119) * fix caching module (hotfix for now) (EleutherAI#2124) * Refactor API models (EleutherAI#2008) * refactor pad_token handling to fn * fix docs * add pad_token_handling to vllm * start on API superclass * don't detokenize the returned logits * streamline vllm tokenizer * add type hint * pre-commit * seems to be in working order * add model to init * refactor api models * nit * cleanup * add pbar * fix type hints * change optional dependencies * json encode chat template * add type hints * deal with different prompt input requiremnts * nits * fix * cache inside async * fix * fix * nits * nits * nits * nit * fixup * fixup * nit * add dummy retry * add dummy retry * handle imports; skip failing test * add type hint * add tests * add dependency to tests * add package names to exception * nit * docs; type hints * handle api key * nit * tokenizer bug * fix tokenizer * nit * nit * add better error messages * nit * remove decorator * CI: install api dep * revert evaluator.py * consolidate * consolidate * nits * nit * fix typealias * nit * nit * nit * Update lm_eval/models/api_models.py typo Co-authored-by: Hailey Schoelkopf <[email protected]> * Update lm_eval/models/openai_completions.py Co-authored-by: Hailey Schoelkopf <[email protected]> * Update lm_eval/models/anthropic_llms.py Co-authored-by: Hailey Schoelkopf <[email protected]> * Update lm_eval/models/api_models.py Co-authored-by: Hailey Schoelkopf <[email protected]> * fix typo * add news section * add info for API * pre-commit * typo * fix bug: unpack logliklehood requests * fix bug: shared gen_kwargs mutated * nit: handle copy properly * Update README.md * Update README.md * Update README.md * Update api_models.py * Update README.md --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * bugfix and docs for API (EleutherAI#2139) * encoding bugfix * encoding bugfix * overload logliklehood rather than loglikehood_tokens * add custom tokenizer * add docs * Update API_guide.md fix link; add note * Update API_guide.md typo * pre-commit * add link in readme * nit * nit * nit * Update API_guide.md nits * Update API_guide.md * Update API_guide.md * Update API_guide.md * Update API_guide.md * Update README.md * Update docs/API_guide.md * Update docs/API_guide.md * Update API_guide.md --------- Co-authored-by: Hailey Schoelkopf <[email protected]> * [Bugfix] add temperature=0 to logprobs and seed args to API models (EleutherAI#2149) * add temperature for log probs * add seed * nit * add new args to test * added warning for api chat models * refactor: limit usage of `scipy` and `skilearn` dependencies (EleutherAI#2097) * refactor: move scipy and sklearn module imports to func imports Signed-off-by: Nathan Weinberg <[email protected]> * refactor: consolidate weighted_f1_score func into lm_eval utils Signed-off-by: Nathan Weinberg <[email protected]> * lint: allow for utils file to have unused imports this allows for shared functions to be defined only once while allowing for the YAML function importing to continue working Signed-off-by: Nathan Weinberg <[email protected]> --------- Signed-off-by: Nathan Weinberg <[email protected]> --------- Signed-off-by: changwangss <[email protected]> Signed-off-by: Elron Bandel <[email protected]> Signed-off-by: elronbandel <[email protected]> Signed-off-by: Nathan Weinberg <[email protected]> Co-authored-by: Nick Doiron <[email protected]> Co-authored-by: Hailey Schoelkopf <[email protected]> Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: zhabuye <[email protected]> Co-authored-by: Edward Gan <[email protected]> Co-authored-by: DongGeon Lee <[email protected]> Co-authored-by: Huazhong Ji <[email protected]> Co-authored-by: Lintang Sutawika <[email protected]> Co-authored-by: Michael Goin <[email protected]> Co-authored-by: jiaqiw09 <[email protected]> Co-authored-by: zhabuye <[email protected]> Co-authored-by: haileyschoelkopf <[email protected]> Co-authored-by: KonradSzafer <[email protected]> Co-authored-by: Nathan Habib <[email protected]> Co-authored-by: Alina Lozovskaia <[email protected]> Co-authored-by: Clémentine Fourrier <[email protected]> Co-authored-by: LSinev <[email protected]> Co-authored-by: anthony-dipofi <[email protected]> Co-authored-by: Harish Vadaparty <[email protected]> Co-authored-by: Maxime <[email protected]> Co-authored-by: MorishT <[email protected]> Co-authored-by: Iker García-Ferrero <[email protected]> Co-authored-by: khalil <[email protected]> Co-authored-by: Zafir Stojanovski <[email protected]> Co-authored-by: Sadra Barikbin <[email protected]> Co-authored-by: Nikita Lozhnikov <[email protected]> Co-authored-by: Baber Abbasi <[email protected]> Co-authored-by: johnwee1 <[email protected]> Co-authored-by: Wang, Chang <[email protected]> Co-authored-by: Yazeed Alnumay <[email protected]> Co-authored-by: Julen Etxaniz <[email protected]> Co-authored-by: achervyakov <[email protected]> Co-authored-by: Stella Biderman <[email protected]> Co-authored-by: jonabur <[email protected]> Co-authored-by: Brendan Murphy <[email protected]> Co-authored-by: Steven Basart <[email protected]> Co-authored-by: Ogundepo Odunayo <[email protected]> Co-authored-by: Nathan Habib <[email protected]> Co-authored-by: Hanwool Albert Lee <[email protected]> Co-authored-by: Choyunhui <[email protected]> Co-authored-by: yhjo <[email protected]> Co-authored-by: Elron Bandel <[email protected]> Co-authored-by: Pankaj Mathur <[email protected]> Co-authored-by: meg <[email protected]> Co-authored-by: Wonung Kim <[email protected]> Co-authored-by: SuperCat <[email protected]> Co-authored-by: Jess <[email protected]> Co-authored-by: Israel Abebe Azime <[email protected]> Co-authored-by: Israel Abebe Azime <[email protected]> Co-authored-by: Nathan Weinberg <[email protected]> Co-authored-by: Ben Shoham Ofir <[email protected]> Co-authored-by: jab13x <[email protected]> Co-authored-by: Jungwhan Kim <[email protected]> Co-authored-by: Jennifer Cwagenberg <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Fixes a bug in which if one passed
--gen_kwargs temperature=0.8
via the CLI and ran a task that usesdo_sample: False
in its default generation kwargs, the temperature setting would be ignored.