aaditya commited on
Commit
c77c689
•
1 Parent(s): 7f2419d

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +2 -5
README.md CHANGED
@@ -17,8 +17,6 @@ model-index:
17
  license: llama3
18
  language:
19
  - en
20
- datasets:
21
- - berkeley-nest/Nectar
22
  widget:
23
  - example_title: OpenBioLLM-70B
24
  messages:
@@ -85,15 +83,14 @@ OpenBioLLM-70B is an advanced open source language model designed specifically f
85
 
86
  🎓 **Superior Performance**: With 70 billion parameters, OpenBioLLM-70B outperforms other open source biomedical language models of similar scale. It has also demonstrated better results compared to larger proprietary & open-source models like GPT-4, Gemini, Meditron-70B, Med-PaLM-1 & Med-PaLM-2 on biomedical benchmarks.
87
 
88
- 🧠 **Advanced Training Techniques**: OpenBioLLM-70B builds upon the powerful foundations of the **Meta-Llama-3-70B-Instruct** and [Meta-Llama-3-70B-Instruct](meta-llama/Meta-Llama-3-70B-Instruct) models. It incorporates the same dataset and fine-tuning recipe as the Starling model, along with a custom diverse medical instruction dataset. Key components of the training pipeline include:
89
 
90
  <div align="center">
91
  <img width="1200px" src="https://cdn-uploads.huggingface.co/production/uploads/5f3fe13d79c1ba4c353d0c19/oPchsJsEpQoGcGXVbh7YS.png">
92
  </div>
93
 
94
 
95
- - **Policy Optimization**: [Fine-Tuning Language Models from Human Preferences (PPO)](https://arxiv.org/abs/1909.08593)
96
- - **Ranking Dataset**: [berkeley-nest/Nectar](https://huggingface.co/datasets/berkeley-nest/Nectar)
97
  - **Fine-tuning dataset**: Custom Medical Instruct dataset (We plan to release a sample training dataset in our upcoming paper; please stay updated)
98
 
99
  This combination of cutting-edge techniques enables OpenBioLLM-70B to align with key capabilities and preferences for biomedical applications.
 
17
  license: llama3
18
  language:
19
  - en
 
 
20
  widget:
21
  - example_title: OpenBioLLM-70B
22
  messages:
 
83
 
84
  🎓 **Superior Performance**: With 70 billion parameters, OpenBioLLM-70B outperforms other open source biomedical language models of similar scale. It has also demonstrated better results compared to larger proprietary & open-source models like GPT-4, Gemini, Meditron-70B, Med-PaLM-1 & Med-PaLM-2 on biomedical benchmarks.
85
 
86
+ 🧠 **Advanced Training Techniques**: OpenBioLLM-70B builds upon the powerful foundations of the **Meta-Llama-3-70B-Instruct** and [Meta-Llama-3-70B-Instruct](meta-llama/Meta-Llama-3-70B-Instruct) models. It incorporates the DPO dataset and fine-tuning recipe along with a custom diverse medical instruction dataset. Key components of the training pipeline include:
87
 
88
  <div align="center">
89
  <img width="1200px" src="https://cdn-uploads.huggingface.co/production/uploads/5f3fe13d79c1ba4c353d0c19/oPchsJsEpQoGcGXVbh7YS.png">
90
  </div>
91
 
92
 
93
+ - **Policy Optimization**: [Direct Preference Optimization: Your Language Model is Secretly a Reward Model (DPO)](https://arxiv.org/abs/2305.18290)
 
94
  - **Fine-tuning dataset**: Custom Medical Instruct dataset (We plan to release a sample training dataset in our upcoming paper; please stay updated)
95
 
96
  This combination of cutting-edge techniques enables OpenBioLLM-70B to align with key capabilities and preferences for biomedical applications.