Add perplexity example to the logprobs user guide
#1071
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
jhills20
reviewed
Feb 27, 2024
There was a problem hiding this comment.
Looks great! Thanks Ankur. Added some small nits. Generally, I think also showing each token's probability would be helpful too, maybe just for one sentence or for both. Because then we could see which tokens the model is more or less confident in, which might help people grasp this intuitively. Could also make prompt to make the sentences briefer if it's too much to do each logprobs for the longer sentences.
Generally good stuff tho! Thanks
examples/Using_logprobs.ipynb
Outdated
| "* Users can easily create a token highlighter using the built in tokenization that comes with enabling `logprobs`. Additionally, the bytes parameter includes the ASCII encoding of each output character, which is particularly useful for reproducing emojis and special characters.\n", | ||
| "\n", | ||
| "4. Calculating perplexity\n", | ||
| "* `logprobs1 can be used to help us assess the model's overall confidence in a result and help us compare the confidence of results from different prompts." |
examples/Using_logprobs.ipynb
Outdated
| "4. Token highlighting and outputting bytes\n", | ||
| "* Users can easily create a token highlighter using the built in tokenization that comes with enabling `logprobs`. Additionally, the bytes parameter includes the ASCII encoding of each output character, which is particularly useful for reproducing emojis and special characters.\n", | ||
| "\n", | ||
| "4. Calculating perplexity\n", |
examples/Using_logprobs.ipynb
Outdated
| "## 5. Conclusion" | ||
| "## 5. Calculating perplexity\n", | ||
| "\n", | ||
| "When looking to assess the model's confidence in a result, it can be useful to calculate perplexity, which is a measure of the uncertainty. Perplexity can be calculated by exponentiating the negative of the average of the logprobs. Generally, a higher perplexity indicates a more uncertain result, and a lower perplexity indicates a more confident result. As such, perplexity can be used to both assess the result of an individual model run and also to helpfully compare the relative confidence of results between model runs. While a high confidence doesn't guarantee result accuracy, it can be a helpful signal that can be paired with other evaluation metrics to build a better understanding of your prompt's behavior.\n", |
There was a problem hiding this comment.
"is a measure of the model's uncertainty."
.."average of the output logprobs"
would remove "helpfully" before compare
examples/Using_logprobs.ipynb
Outdated
| "* Users can easily create a token highlighter using the built in tokenization that comes with enabling `logprobs`. Additionally, the bytes parameter includes the ASCII encoding of each output character, which is particularly useful for reproducing emojis and special characters.\n", | ||
| "\n", | ||
| "4. Calculating perplexity\n", | ||
| "* `logprobs1 can be used to help us assess the model's overall confidence in a result and help us compare the confidence of results from different prompts." |
There was a problem hiding this comment.
nit - fix logprobs formatting
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.
Summary
Adds a section to demonstrate how logprobs can be used to assess model confidence in overall results. The changes introduce the concept of perplexity, add code to calculate it for two examples, and display the output.
Motivation
These changes build on a suggestion at the end of the original guide to use logprobs for different evaluation metrics, filling in a recommended extension.
For new content
When contributing new content, read through our contribution guidelines, and mark the following action items as completed:
We will rate each of these areas on a scale from 1 to 4, and will only accept contributions that score 3 or higher on all areas. Refer to our contribution guidelines for more details.