Add Loss Logic Eval (openai#82)

# Thank you for contributing an eval! ♥️

🚨 Please make sure your PR follows these guidelines, __failure to follow
the guidelines below will result in the PR being closed automatically__.
Note that even if the criteria are met, that does not guarantee the PR
will be merged nor GPT-4 access granted. 🚨

__PLEASE READ THIS__:

In order for a PR to be merged, it must fail on GPT-4. We are aware that
right now, users do not have access, so you will not be able to tell if
the eval fails or not. Please run your eval with GPT-3.5-Turbo, but keep
in mind as we run the eval, if GPT-4 gets higher than 90% on the eval,
we will likely reject since GPT-4 is already capable of completing the
task.

We plan to roll out a way for users submitting evals to see the eval
performance on GPT-4 soon. Stay tuned! Until then, you will not be able
to see the eval performance on GPT-4.

## Eval details 📑
### Eval name
Loss Logic

### Eval description

A store bought a watermelon for $5 and sold it for a different price,
receiving counterfeit money in the transaction. They also had to give
change to the buyer. The net loss for the store varies based on the
specific details of the transaction.

### What makes this a useful eval?

* Tests comprehension and problem-solving skills: The scenarios provided
require the AI to understand and analyze the given information to
determine the net loss for the store.

* Addresses real-world situations: Counterfeit money transactions are a
real concern for businesses. This eval allows the AI to demonstrate its
understanding of financial transactions and the impact of counterfeit
money on a store's net loss.


## Criteria for a good eval ✅

Below are some of the criteria we look for in a good eval. In general,
we are seeking cases where the model does not do a good job despite
being capable of generating a good response (note that there are some
things large language models cannot do, so those would not make good
evals).

Your eval should be:

- [x] Thematically consistent: The eval should be thematically
consistent. We'd like to see a number of prompts all demonstrating some
particular failure mode. For example, we can create an eval on cases
where the model fails to reason about the physical world.
- [x] Contains failures where a human can do the task, but either GPT-4
or GPT-3.5-Turbo could not.
- [x] Includes good signal around what is the right behavior. This means
either a correct answer for `Basic` evals or the `Fact` Model-graded
eval, or an exhaustive rubric for evaluating answers for the `Criteria`
Model-graded eval.
- [ ] Include at least 100 high quality examples

If there is anything else that makes your eval worth including, please
document it below.

### Unique eval value

Involving the judgement of the truth or falsehood of real-world things,
which is a capability that GPT currently lacks.

## Eval structure 🏗️

Your eval should
- [x] Check that your data is in `evals/registry/data/{name}`
- [x] Check that your yaml is registered at
`evals/registry/evals/{name}.jsonl`
- [x] Ensure you have the right to use the data you submit via this eval

(For now, we will only be approving evals that use one of the existing
eval classes. You may still write custom eval classes for your own
cases, and we may consider merging them in the future.)

## Final checklist 👀

### Submission agreement

By contributing to Evals, you are agreeing to make your evaluation logic
and data under the same MIT license as this repository. You must have
adequate rights to upload any data used in an Eval. OpenAI reserves the
right to use this data in future service improvements to our product.
Contributions to OpenAI Evals will be subject to our usual Usage
Policies (https://platform.openai.com/docs/usage-policies).

- [x] I agree that my submission will be made available under an MIT
license and complies with OpenAI's usage policies.

### Email address validation

If your submission is accepted, we will be granting GPT-4 access to a
limited number of contributors. Access will be given to the email
address associated with the merged pull request.

- [x] I acknowledge that GPT-4 access will only be granted, if
applicable, to the email address used for my merged pull request.

### Limited availability acknowledgement

We know that you might be excited to contribute to OpenAI's mission,
help improve our models, and gain access to GPT-4. However, due to the
requirements mentioned above and high volume of submissions, we will not
be able to accept all submissions and thus not grant everyone who opens
a PR GPT-4 access. We know this is disappointing, but we hope to set the
right expectation before you open this PR.

- [x] I understand that opening a PR, even if it meets the requirements
above, does not guarantee the PR will be merged nor GPT-4 access
granted.

### Submit eval

- [x] I have filled out all required fields in the evals PR form
- [x] (Ignore if not submitting code) I have run `pip install
pre-commit; pre-commit install` and have verified that `black`, `isort`,
and `autoflake` are running when I commit and push

Failure to fill out all required fields will result in the PR being
closed.

### Eval JSON data 

Since we are using Git LFS, we are asking eval submitters to add in as
many Eval Samples (at least 5) from their contribution here:

<details>
  <summary>View evals in JSON</summary>

  ### Eval
  ```jsonl
{"input":[{"role":"system","content":"The store purchased a watermelon
for $5 and sold it for $7. Unfortunately, they received $10 in
counterfeit money after the transaction and had to give the buyer $3 in
change. What was the net loss for the store? Let's think step by
step:"}],"ideal":"Sure, let's break down the transaction step by step to
determine the net loss for the store: The store purchases a watermelon
for $5. The store sells the watermelon for $7. This would have been a $2
profit if the money received was real. The store receives $10 in
counterfeit money. This money is worthless, so the store doesn't
actually receive any value from the sale. The store gives the buyer $3
in change. This is a real loss, as the store loses $3 in real money.
Now, let's calculate the net loss: The store spent $5 to purchase the
watermelon. The store lost $3 when giving the buyer change. So the net
loss for the store is $5 (cost of watermelon) + $3 (change given) =
$8."}
{"input":[{"role":"system","content":"The store bought a watermelon for
$5 but sold it for $4. Furthermore, they received $10 in counterfeit
money from the buyer and had to provide $6 in change. What was the net
loss for the store? Let's consider the process in
detail:"}],"ideal":"Sure, Let's think step by step: The store purchased
a watermelon for $5. The store sold the watermelon for $4, which means
they incurred a loss of $1 on the sale. The buyer gave the store $10 in
counterfeit money, which the store accepted during the transaction. The
store gave the buyer $6 in change. Since the store accepted $10 in
counterfeit money, they lost $10 in that transaction. The net loss for
the store is the sum of the loss on the watermelon sale and the loss
from accepting the counterfeit money, which is $1 + $10 = $11.
Therefore, the net loss for the store is $11."}
{"input":[{"role":"system","content":"The store bought a watermelon for
$5 and sold it for the same price of $5. However, they received $10 in
counterfeit money from the buyer and had to give the buyer $5 in change.
What was the net loss for the store? Let's consider the process step by
step:"}],"ideal":"Step 1: The store purchases a watermelon for $5.
Store's expenses: $5 Step 2: The store sells the watermelon for $5 but
receives $10 in counterfeit money. The counterfeit money has no value,
so the store effectively received nothing for the watermelon. Store's
revenue: $0 Step 3: The store gives the buyer $5 in change. Since the
money given in change is real, this is an additional expense for the
store. Store's expenses: $5 (watermelon) + $5 (change) = $10 Step 4:
Calculate the net loss. Net loss = Expenses - Revenue Net loss = $10
(expenses) - $0 (revenue) Net loss = $10 The store's net loss for this
transaction is $10."}
{"input":[{"role":"system","content":"The store bought a watermelon for
$5 and sold it for $7. Regrettably, they received $7 in counterfeit
money from the buyer after the transaction. What was the net loss for
the store? Let's consider the process step by step:"}],"ideal":"Sure,
let's break down the transaction step by step: The store purchased a
watermelon for $5. They sold the watermelon for $7, but received
counterfeit money. To determine the net loss, let's analyze the costs
and the revenue: Costs: The store spent $5 on the watermelon. Revenue:
The store received $7 in counterfeit money, which has no value, so their
revenue is $0. Net loss: To calculate the net loss, subtract the revenue
from the costs: $5 (costs) - $0 (revenue) = $5. So, the net loss for the
store is $5."}
{"input":[{"role":"system","content":"The store was given a
complimentary watermelon, which they then sold for $7. Regrettably,
following the transaction, they received $10 in fake money and had to
provide the buyer with $3 in change. Let's consider the following steps
in determining the store's net loss:"}],"ideal":"To determine the
store's net loss, we can consider the following steps: Assess the value
of the complimentary watermelon: Since the watermelon was given to the
store for free, it didn't cost them anything. Therefore, the store's
initial cost for the watermelon is $0. Calculate the revenue from
selling the watermelon: The store sold the watermelon for $7. However,
they received $10 in fake money, which has no value, so the actual
revenue is $0. Determine the cost of the change provided: Since the
store provided the buyer with $3 in change, this is an additional cost
to the store. Calculate the net loss: Subtract the revenue (Step 2) from
the sum of the initial cost (Step 1) and the cost of the change (Step
3). In this case: Net loss = (Initial cost + Cost of change) - Revenue
Net loss = ($0 + $3) - $0 Net loss = $3 The store's net loss from this
transaction is $3."}
  ```
</details>

evals/registry/evals/loss-logic.yaml

@@ -0,0 +1,9 @@
+loss-logic-fact:
+ id: loss-logic-fact.dev.v0
+ metrics: [accuracy]
+loss-logic-fact.dev.v0:
+ class: evals.elsuite.modelgraded.classify:ModelBasedClassify
+ args:
+ samples_jsonl: loss_logic/samples.jsonl
+ eval_type: cot_classify
+ modelgraded_spec_file: fact