Update text_generation_utils.py

[StellaAthena]

Fix off-by-one-error per #627

[fpgaminer]

Works correctly on my setup! Using this config:

    "maximum_tokens": 64,
    "temperature": 0.7,
    "top_k": 40,
    "seed": 1235,
    "text-gen-type": "interactive",

And typing "Hello" into the interactive prompt, main on 6a76fe6a8011d6c9b9d44e55a1ade59755dd0435 incorrectly generates:

. I’m a writer, and this is my blog. My goal is to inspire you to write and to make writing fun. I’m from Ohio, love to do laundry, and can’t imagine ever owning a dishwasher. My husband would say I’m kind of crazy. I’

Whereas this patch, 765d8f34fd20ec2e95e9f69c3bc73a3822a5b506, correctly generates:

. I’m a writer, and this is my blog. My goal is to inspire you to write and to make writing fun. I’m from Ohio, love to do laundry, and can’t imagine ever owning a dishwasher. My husband would say I’m kind of crazy. I’m

(Notice the inclusion of the last token, "m", which was dropped by the buggy code.)

So, looks good.

[StellaAthena]

@EricHallahan can you review + merge?