Update README.md
Browse files
README.md
CHANGED
@@ -59,16 +59,18 @@ tags:
|
|
59 |
# Democratizing Medical LLMs For Much More Languages
|
60 |
|
61 |
Covering 12 Major Languages including English, Chinese, French, Hindi, Spanish, Arabic, Russian, Japanese, Korean, German, Italian, Portuguese and 38 Minor Languages So far.
|
62 |
-
<center>
|
63 |
|
64 |
|
65 |
|
66 |
<p align="center">
|
67 |
-
๐ <a href="https://arxiv.org/abs/2410.10626" target="_blank">Paper</a> โข ๐ <a href="" target="_blank">Demo</a> โข ๐ค <a href="https://huggingface.co/datasets/FreedomIntelligence/ApolloMoEDataset" target="_blank">ApolloMoEDataset</a> โข ๐ค <a href="https://huggingface.co/datasets/FreedomIntelligence/ApolloMoEBench" target="_blank">ApolloMoEBench</a> โข ๐ค <a href="https://huggingface.co/collections/FreedomIntelligence/apollomoe-and-apollo2-670ddebe3bb1ba1aebabbf2c" target="_blank">Models</a> โข ๐ <a href="https://github.com/FreedomIntelligence/
|
68 |
</p>
|
69 |
|
|
|
|
|
70 |
![Apollo](assets/apollo_medium_final.png)
|
71 |
|
|
|
72 |
## ๐ Update
|
73 |
|
74 |
* **[2024.10.15]** ApolloMoE repo is published๏ผ๐
|
@@ -90,42 +92,47 @@ Covering 12 Major Languages including English, Chinese, French, Hindi, Spanish,
|
|
90 |
<details>
|
91 |
<summary>Click to view the MoE routing image</summary>
|
92 |
|
93 |
-
![ApolloMoE](
|
94 |
|
95 |
</details>
|
96 |
|
97 |
## Results
|
98 |
|
99 |
-
|
100 |
-
๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo2-0.5B" target="_blank">Apollo2-0.5B</a>
|
101 |
|
|
|
|
|
102 |
<details>
|
103 |
<summary>Click to view the Dense Models Results</summary>
|
104 |
-
|
105 |
![ApolloMoE](assets/dense_results.png)
|
106 |
|
107 |
</details>
|
108 |
|
109 |
-
|
|
|
110 |
๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo-MoE-0.5B" target="_blank">Apollo-MoE-0.5B</a> โข ๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo-MoE-1.5B" target="_blank">Apollo-MoE-1.5B</a> โข ๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo-MoE-7B" target="_blank">Apollo-MoE-7B</a>
|
111 |
-
|
112 |
<details>
|
113 |
<summary>Click to view the Post-MoE Models Results</summary>
|
114 |
-
|
115 |
![ApolloMoE](assets/post_moe_results.png)
|
116 |
|
117 |
</details>
|
118 |
|
|
|
|
|
119 |
|
120 |
## Usage Format
|
121 |
-
|
122 |
- 0.5B, 1.5B, 7B: User:{query}\nAssistant:{response}<|endoftext|>
|
123 |
- 2B, 9B: User:{query}\nAssistant:{response}\<eos\>
|
124 |
- 3.8B: <|user|>\n{query}<|end|><|assisitant|>\n{response}<|end|>
|
125 |
|
126 |
-
|
127 |
- 0.5B, 1.5B, 7B: User:{query}\nAssistant:{response}<|endoftext|>
|
128 |
-
|
129 |
## Dataset & Evaluation
|
130 |
|
131 |
- Dataset
|
@@ -139,12 +146,12 @@ Covering 12 Major Languages including English, Chinese, French, Hindi, Spanish,
|
|
139 |
|
140 |
|
141 |
</details>
|
142 |
-
|
143 |
- Evaluation
|
144 |
๐ค <a href="https://huggingface.co/datasets/FreedomIntelligence/ApolloMoEBench" target="_blank">ApolloMoEBench</a>
|
145 |
|
146 |
<details><summary>Click to expand</summary>
|
147 |
-
|
148 |
- EN:
|
149 |
- [MedQA-USMLE](https://huggingface.co/datasets/GBaker/MedQA-USMLE-4-options)
|
150 |
- [MedMCQA](https://huggingface.co/datasets/medmcqa/viewer/default/test)
|
@@ -180,17 +187,16 @@ Covering 12 Major Languages including English, Chinese, French, Hindi, Spanish,
|
|
180 |
- PT: [BioInstructQA](https://huggingface.co/datasets/BioMistral/BioInstructQA): Portuguese part
|
181 |
- RU: [RuMedBench](https://github.com/sb-ai-lab/MedBench)
|
182 |
|
183 |
-
|
184 |
-
|
185 |
|
186 |
|
187 |
</details>
|
188 |
|
189 |
-
|
190 |
## Results reproduction
|
191 |
<details><summary>Click to expand</summary>
|
192 |
|
193 |
-
|
194 |
We take Apollo2-7B or Apollo-MoE-0.5B as example
|
195 |
1. Download Dataset for project:
|
196 |
|
|
|
59 |
# Democratizing Medical LLMs For Much More Languages
|
60 |
|
61 |
Covering 12 Major Languages including English, Chinese, French, Hindi, Spanish, Arabic, Russian, Japanese, Korean, German, Italian, Portuguese and 38 Minor Languages So far.
|
|
|
62 |
|
63 |
|
64 |
|
65 |
<p align="center">
|
66 |
+
๐ <a href="https://arxiv.org/abs/2410.10626" target="_blank">Paper</a> โข ๐ <a href="" target="_blank">Demo</a> โข ๐ค <a href="https://huggingface.co/datasets/FreedomIntelligence/ApolloMoEDataset" target="_blank">ApolloMoEDataset</a> โข ๐ค <a href="https://huggingface.co/datasets/FreedomIntelligence/ApolloMoEBench" target="_blank">ApolloMoEBench</a> โข ๐ค <a href="https://huggingface.co/collections/FreedomIntelligence/apollomoe-and-apollo2-670ddebe3bb1ba1aebabbf2c" target="_blank">Models</a> โข๐ <a href="https://github.com/FreedomIntelligence/Apollo" target="_blank">Apollo</a> โข ๐ <a href="https://github.com/FreedomIntelligence/ApolloMoE" target="_blank">ApolloMoE</a>
|
67 |
</p>
|
68 |
|
69 |
+
|
70 |
+
|
71 |
![Apollo](assets/apollo_medium_final.png)
|
72 |
|
73 |
+
|
74 |
## ๐ Update
|
75 |
|
76 |
* **[2024.10.15]** ApolloMoE repo is published๏ผ๐
|
|
|
92 |
<details>
|
93 |
<summary>Click to view the MoE routing image</summary>
|
94 |
|
95 |
+
![ApolloMoE](assets/hybrid_routing.png)
|
96 |
|
97 |
</details>
|
98 |
|
99 |
## Results
|
100 |
|
101 |
+
#### Dense
|
102 |
+
๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo2-0.5B" target="_blank">Apollo2-0.5B</a> โข ๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo2-1.5B" target="_blank">Apollo2-1.5B</a> โข ๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo2-2B" target="_blank">Apollo2-2B</a>
|
103 |
|
104 |
+
๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo2-3.8B" target="_blank">Apollo2-3.8B</a> โข ๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo2-7B" target="_blank">Apollo2-7B</a> โข ๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo2-9B" target="_blank">Apollo2-9B</a>
|
105 |
+
|
106 |
<details>
|
107 |
<summary>Click to view the Dense Models Results</summary>
|
108 |
+
|
109 |
![ApolloMoE](assets/dense_results.png)
|
110 |
|
111 |
</details>
|
112 |
|
113 |
+
|
114 |
+
#### Post-MoE
|
115 |
๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo-MoE-0.5B" target="_blank">Apollo-MoE-0.5B</a> โข ๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo-MoE-1.5B" target="_blank">Apollo-MoE-1.5B</a> โข ๐ค <a href="https://huggingface.co/FreedomIntelligence/Apollo-MoE-7B" target="_blank">Apollo-MoE-7B</a>
|
116 |
+
|
117 |
<details>
|
118 |
<summary>Click to view the Post-MoE Models Results</summary>
|
119 |
+
|
120 |
![ApolloMoE](assets/post_moe_results.png)
|
121 |
|
122 |
</details>
|
123 |
|
124 |
+
|
125 |
+
|
126 |
|
127 |
## Usage Format
|
128 |
+
##### Apollo2
|
129 |
- 0.5B, 1.5B, 7B: User:{query}\nAssistant:{response}<|endoftext|>
|
130 |
- 2B, 9B: User:{query}\nAssistant:{response}\<eos\>
|
131 |
- 3.8B: <|user|>\n{query}<|end|><|assisitant|>\n{response}<|end|>
|
132 |
|
133 |
+
##### Apollo-MoE
|
134 |
- 0.5B, 1.5B, 7B: User:{query}\nAssistant:{response}<|endoftext|>
|
135 |
+
|
136 |
## Dataset & Evaluation
|
137 |
|
138 |
- Dataset
|
|
|
146 |
|
147 |
|
148 |
</details>
|
149 |
+
|
150 |
- Evaluation
|
151 |
๐ค <a href="https://huggingface.co/datasets/FreedomIntelligence/ApolloMoEBench" target="_blank">ApolloMoEBench</a>
|
152 |
|
153 |
<details><summary>Click to expand</summary>
|
154 |
+
|
155 |
- EN:
|
156 |
- [MedQA-USMLE](https://huggingface.co/datasets/GBaker/MedQA-USMLE-4-options)
|
157 |
- [MedMCQA](https://huggingface.co/datasets/medmcqa/viewer/default/test)
|
|
|
187 |
- PT: [BioInstructQA](https://huggingface.co/datasets/BioMistral/BioInstructQA): Portuguese part
|
188 |
- RU: [RuMedBench](https://github.com/sb-ai-lab/MedBench)
|
189 |
|
190 |
+
|
|
|
191 |
|
192 |
|
193 |
</details>
|
194 |
|
195 |
+
|
196 |
## Results reproduction
|
197 |
<details><summary>Click to expand</summary>
|
198 |
|
199 |
+
|
200 |
We take Apollo2-7B or Apollo-MoE-0.5B as example
|
201 |
1. Download Dataset for project:
|
202 |
|