Update README.md
Browse files
README.md
CHANGED
@@ -8,21 +8,20 @@ language:
|
|
8 |
# EvoLLM-v1-JP-7B
|
9 |
|
10 |
<!-- Provide a quick summary of what the model is/does. -->
|
11 |
-
EvoLLM-v1-JP-7B is a evolved Japanese Math LLM.
|
12 |
|
13 |
## Model Details
|
14 |
|
15 |
### Model Description
|
16 |
|
17 |
<!-- Provide a longer summary of what this model is. -->
|
18 |
-
|
19 |
-
This is the model card of a 🤗 transformers model that has been pushed on the Hub. This model card has been automatically generated.
|
20 |
|
21 |
- **Developed by:** [Sakana AI](https://sakana.ai/)
|
22 |
- **Model type:** Autoregressive Language Model
|
23 |
- **Language(s):** Japanese
|
24 |
- **License:** [MICROSOFT RESEARCH LICENSE TERMS](./LICENSE)
|
25 |
-
- **
|
26 |
- [augmxnt/shisa-gamma-7b-v1](https://huggingface.co/augmxnt/shisa-gamma-7b-v1)
|
27 |
- [WizardLM/WizardMath-7B-V1.1](https://huggingface.co/WizardLM/WizardMath-7B-V1.1)
|
28 |
- [GAIR/Abel-7B-002](https://huggingface.co/GAIR/Abel-7B-002)
|
@@ -72,12 +71,21 @@ print(generated_text)
|
|
72 |
|
73 |
## Evaluation
|
74 |
|
75 |
-
|
|
|
|
|
76 |
|
77 |
|
78 |
## Citation
|
79 |
|
80 |
```bibtex
|
81 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
82 |
```
|
83 |
|
|
|
8 |
# EvoLLM-v1-JP-7B
|
9 |
|
10 |
<!-- Provide a quick summary of what the model is/does. -->
|
11 |
+
EvoLLM-v1-JP-7B is a evolved Japanese Math LLM.
|
12 |
|
13 |
## Model Details
|
14 |
|
15 |
### Model Description
|
16 |
|
17 |
<!-- Provide a longer summary of what this model is. -->
|
18 |
+
EvoLLM-v1-JP-7B is a Japanese Math LLM, merged the following source models in the Parameter Space (PS) by using an evolutionary approach.
|
|
|
19 |
|
20 |
- **Developed by:** [Sakana AI](https://sakana.ai/)
|
21 |
- **Model type:** Autoregressive Language Model
|
22 |
- **Language(s):** Japanese
|
23 |
- **License:** [MICROSOFT RESEARCH LICENSE TERMS](./LICENSE)
|
24 |
+
- **Source models:**
|
25 |
- [augmxnt/shisa-gamma-7b-v1](https://huggingface.co/augmxnt/shisa-gamma-7b-v1)
|
26 |
- [WizardLM/WizardMath-7B-V1.1](https://huggingface.co/WizardLM/WizardMath-7B-V1.1)
|
27 |
- [GAIR/Abel-7B-002](https://huggingface.co/GAIR/Abel-7B-002)
|
|
|
71 |
|
72 |
## Evaluation
|
73 |
|
74 |
+
We present the results that compares the performance of the our evolved LLMs compared to the source LLMs. To reproduce the results, please use [our Github repository](https://github.com/SakanaAI/evolving-merged-models).
|
75 |
+
|
76 |
+
![eval-results](./evollm-math-results.png)
|
77 |
|
78 |
|
79 |
## Citation
|
80 |
|
81 |
```bibtex
|
82 |
+
@misc{sakana2024evofactory,
|
83 |
+
title = {Evolutionary Optimization of Model Merging Recipes},
|
84 |
+
author. = {Takuya Akiba and Makoto Shing and Yujin Tang and Qi Sun and David Ha},
|
85 |
+
year = {2024},
|
86 |
+
eprint = {TODO},
|
87 |
+
archivePrefix = {arXiv},
|
88 |
+
primaryClass = {cs.CV}
|
89 |
+
}
|
90 |
```
|
91 |
|