yuxiang630 commited on
Commit
cc7bf2c
1 Parent(s): b120b50

refactor: use hyperlinks for images

Browse files
Files changed (1) hide show
  1. README.md +6 -7
README.md CHANGED
@@ -86,19 +86,17 @@ model-index:
86
 
87
  # StarCoder2-Instruct: Self-Aligned, Transparent, and Fully Permissive
88
 
89
- <!-- <center>
90
- <img src="https://huggingface.co/datasets/bigcode/admin_private/resolve/main/starcoder2_banner.png" alt="SC2" width="900" height="600">
91
- </center> -->
92
 
93
  ## Model Summary
94
 
95
  We introduce StarCoder2-15B-Instruct-v0.1, the very first entirely self-aligned code Large Language Model (LLM) trained with a fully permissive and transparent pipeline. Our open-source pipeline uses StarCoder2-15B to generate thousands of instruction-response pairs, which are then used to fine-tune StarCoder-15B itself without any human annotations or distilled data from huge and proprietary LLMs.
96
 
97
- - **Model:** [bigcode/starCoder2-15b-instruct-v0.1](https://huggingface.co/bigcode/starcoder2-instruct-15b-v0.1)
98
  - **Code:** [bigcode-project/starcoder2-self-align](https://github.com/bigcode-project/starcoder2-self-align)
99
  - **Dataset:** [bigcode/self-oss-instruct-sc2-exec-filter-50k](https://huggingface.co/datasets/bigcode/self-oss-instruct-sc2-exec-filter-50k/)
100
 
101
- ![self-alignment pipeline](assets/star-align-pipeline.svg)
102
 
103
  ## Use
104
 
@@ -177,14 +175,15 @@ The model also inherits the bias, risks, and limitations from its base StarCoder
177
 
178
  ## Evaluation on EvalPlus, LiveCodeBench, and DS-1000
179
 
180
- ![EvalPlus](assets/sc2-instruct-evalplus.png)
181
 
182
- ![LiveCodeBench and DS-1000](assets/sc2-instruct-lcb-ds.png)
183
 
184
  ## Training Details
185
 
186
  ### Hyperparameters
187
 
 
188
  - **Learning rate:** 1e-5
189
  - **Epoch:** 4
190
  - **Batch size:** 64
 
86
 
87
  # StarCoder2-Instruct: Self-Aligned, Transparent, and Fully Permissive
88
 
89
+ ![Banner](https://huggingface.co/datasets/bigcode/starcoder2-instruct-assets/resolve/main/banner.png)
 
 
90
 
91
  ## Model Summary
92
 
93
  We introduce StarCoder2-15B-Instruct-v0.1, the very first entirely self-aligned code Large Language Model (LLM) trained with a fully permissive and transparent pipeline. Our open-source pipeline uses StarCoder2-15B to generate thousands of instruction-response pairs, which are then used to fine-tune StarCoder-15B itself without any human annotations or distilled data from huge and proprietary LLMs.
94
 
95
+ - **Model:** [bigcode/starcoder2-15b-instruct-v0.1](https://huggingface.co/bigcode/starcoder2-instruct-15b-v0.1)
96
  - **Code:** [bigcode-project/starcoder2-self-align](https://github.com/bigcode-project/starcoder2-self-align)
97
  - **Dataset:** [bigcode/self-oss-instruct-sc2-exec-filter-50k](https://huggingface.co/datasets/bigcode/self-oss-instruct-sc2-exec-filter-50k/)
98
 
99
+ ![self-alignment pipeline](https://huggingface.co/datasets/bigcode/starcoder2-instruct-assets/resolve/main/method.png)
100
 
101
  ## Use
102
 
 
175
 
176
  ## Evaluation on EvalPlus, LiveCodeBench, and DS-1000
177
 
178
+ ![EvalPlus](https://huggingface.co/datasets/bigcode/starcoder2-instruct-assets/resolve/main/evalplus.png)
179
 
180
+ ![LiveCodeBench and DS-1000](https://huggingface.co/datasets/bigcode/starcoder2-instruct-assets/resolve/main/lcb-ds1000.png)
181
 
182
  ## Training Details
183
 
184
  ### Hyperparameters
185
 
186
+ - **Optimizer:** Adafactor
187
  - **Learning rate:** 1e-5
188
  - **Epoch:** 4
189
  - **Batch size:** 64