gugarosa commited on
Commit
07d9363
1 Parent(s): 03b9f69

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +10 -1
README.md CHANGED
@@ -1,5 +1,4 @@
1
  ---
2
- inference: false
3
  license: mit
4
  license_link: https://huggingface.co/microsoft/phi-1/resolve/main/LICENSE
5
  language:
@@ -13,6 +12,16 @@ tags:
13
 
14
  The language model Phi-1 is a Transformer with 1.3 billion parameters, specialized for basic Python coding. Its training involved a variety of data sources, including subsets of Python codes from [The Stack v1.2](https://huggingface.co/datasets/bigcode/the-stack), Q&A content from [StackOverflow](https://archive.org/download/stackexchange), competition code from [code_contests](https://github.com/deepmind/code_contests), and synthetic Python textbooks and exercises generated by [gpt-3.5-turbo-0301](https://platform.openai.com/docs/models/gpt-3-5). Even though the model and the datasets are relatively small compared to contemporary Large Language Models (LLMs), Phi-1 has demonstrated an impressive accuracy rate exceeding 50% on the simple Python coding benchmark, HumanEval.
15
 
 
 
 
 
 
 
 
 
 
 
16
  ## Intended Uses
17
 
18
  Given the nature of the training data, Phi-1 is best suited for prompts using the code format:
 
1
  ---
 
2
  license: mit
3
  license_link: https://huggingface.co/microsoft/phi-1/resolve/main/LICENSE
4
  language:
 
12
 
13
  The language model Phi-1 is a Transformer with 1.3 billion parameters, specialized for basic Python coding. Its training involved a variety of data sources, including subsets of Python codes from [The Stack v1.2](https://huggingface.co/datasets/bigcode/the-stack), Q&A content from [StackOverflow](https://archive.org/download/stackexchange), competition code from [code_contests](https://github.com/deepmind/code_contests), and synthetic Python textbooks and exercises generated by [gpt-3.5-turbo-0301](https://platform.openai.com/docs/models/gpt-3-5). Even though the model and the datasets are relatively small compared to contemporary Large Language Models (LLMs), Phi-1 has demonstrated an impressive accuracy rate exceeding 50% on the simple Python coding benchmark, HumanEval.
14
 
15
+ ## How to Use
16
+
17
+ Phi-1 has been integrated in the development version (4.37.0.dev) of `transformers`. Until the official version is released through `pip`, ensure that you are doing one of the following:
18
+
19
+ * When loading the model, ensure that `trust_remote_code=True` is passed as an argument of the `from_pretrained()` function.
20
+
21
+ * Update your local `transformers` to the development version: `pip uninstall -y transformers && pip install git+https://github.com/huggingface/transformers`. The previous command is an alternative to cloning and installing from the source.
22
+
23
+ The current `transformers` version can be verified with: `pip list | grep transformers`.
24
+
25
  ## Intended Uses
26
 
27
  Given the nature of the training data, Phi-1 is best suited for prompts using the code format: