TheBloke commited on
Commit
59b4e75
·
1 Parent(s): 06ee1d1

Add README

Browse files
Files changed (1) hide show
  1. README.md +46 -0
README.md CHANGED
@@ -1,3 +1,49 @@
1
  ---
2
  license: other
3
  ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: other
3
  ---
4
+
5
+ # Koala: A Dialogue Model for Academic Research
6
+ This repo contains the weights of the Koala 7B model produced at Berkeley. It is the result of combining the diffs from https://huggingface.co/young-geng/koala with the original Llama 7B model.
7
+
8
+ This version has then been converted to HF format.
9
+
10
+ The following commands were run to produce this repo:
11
+ ```
12
+ git clone https://github.com/young-geng/EasyLM
13
+
14
+ git clone https://huggingface.co/nyanko7/LLaMA-7B
15
+
16
+ git clone https://huggingface.co/young-geng/koala koala_diffs
17
+
18
+ PYTHON_PATH="${PWD}:$PYTHONPATH" python \
19
+ -m EasyLM.models.llama.convert_torch_to_easylm \
20
+ --checkpoint_dir=/content/LLaMA-7B \
21
+ --output_file=/content/llama-7B-LM \
22
+ --streaming=True
23
+
24
+ PYTHON_PATH="${PWD}:$PYTHONPATH" python \
25
+ -m EasyLM.scripts.diff_checkpoint --recover_diff=True \
26
+ --load_base_checkpoint='params::/content/llama-7B-LM' \
27
+ --load_target_checkpoint='params::/content/koala_diffs/koala_7b_diff_v2' \
28
+ --output_file=/content/koala_7b.diff.weights \
29
+ --streaming=True
30
+
31
+ PYTHON_PATH="${PWD}:$PYTHONPATH" python \
32
+ -m EasyLM.models.llama.convert_easylm_to_hf --model_size=7b \
33
+ --output_dir=/content/koala-7B-HF \
34
+ --load_checkpoint='params::/content/koala_7b.diff.weights' \
35
+ --tokenizer_path=/content/LLaMA-7B/tokenizer.model
36
+ ```
37
+
38
+ Check out the following links to learn more about the Berkeley Koala model.
39
+ * [Blog post](https://bair.berkeley.edu/blog/2023/04/03/koala/)
40
+ * [Online demo](https://koala.lmsys.org/)
41
+ * [EasyLM: training and serving framework on GitHub](https://github.com/young-geng/EasyLM)
42
+ * [Documentation for running Koala locally](https://github.com/young-geng/EasyLM/blob/main/docs/koala.md)
43
+
44
+ ## License
45
+ The model weights are intended for academic research only, subject to the
46
+ [model License of LLaMA](https://github.com/facebookresearch/llama/blob/main/MODEL_CARD.md),
47
+ [Terms of Use of the data generated by OpenAI](https://openai.com/policies/terms-of-use),
48
+ and [Privacy Practices of ShareGPT](https://chrome.google.com/webstore/detail/sharegpt-share-your-chatg/daiacboceoaocpibfodeljbdfacokfjb).
49
+ Any other usage of the model weights, including but not limited to commercial usage, is strictly prohibited.