OllieStanley
commited on
Commit
•
03ec016
1
Parent(s):
d82939e
Update README.md
Browse files
README.md
CHANGED
@@ -2,9 +2,9 @@
|
|
2 |
license: other
|
3 |
---
|
4 |
|
5 |
-
# OpenAssistant
|
6 |
|
7 |
-
Due to the license attached to
|
8 |
|
9 |
Thanks to Mick for writing the `xor_codec.py` script which enables this process
|
10 |
|
@@ -12,9 +12,13 @@ Thanks to Mick for writing the `xor_codec.py` script which enables this process
|
|
12 |
|
13 |
Note: This process applies to `oasst-rlhf-2-llama-30b-7k-steps` model. The same process can be applied to other models in future, but the checksums will be different..
|
14 |
|
15 |
-
|
16 |
|
17 |
-
|
|
|
|
|
|
|
|
|
18 |
|
19 |
```
|
20 |
f856e9d99c30855d6ead4d00cc3a5573 consolidated.00.pth
|
@@ -24,46 +28,101 @@ ea0405cdb5bc638fee12de614f729ebc consolidated.03.pth
|
|
24 |
4babdbd05b8923226a9e9622492054b6 params.json
|
25 |
```
|
26 |
|
27 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
28 |
|
29 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
30 |
|
31 |
```
|
32 |
-
|
33 |
```
|
34 |
|
35 |
-
Run `find -type f -exec md5sum "{}"
|
36 |
|
37 |
```
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
f845ecc481cb92b8a0586c2ce288b828 ./pytorch_model-00004-of-00007.bin
|
42 |
-
f3b13d089840e6caf22cd6dd05b77ef0 ./pytorch_model-00005-of-00007.bin
|
43 |
-
12e0d2d7a9c00c4237b1b0143c48a05e ./pytorch_model-00007-of-00007.bin
|
44 |
-
1348f7c8bb3ee4408b69305a10bdfafb ./pytorch_model-00006-of-00007.bin
|
45 |
aee09e21813368c49baaece120125ae3 ./generation_config.json
|
|
|
|
|
46 |
eeec4125e9c7560836b4873b6f8e3025 ./tokenizer.model
|
|
|
47 |
598538f18fed1877b41f77de034c0c8a ./config.json
|
48 |
fdb311c39b8659a5d5c1991339bafc09 ./tokenizer.json
|
49 |
-
|
50 |
edd1a5897748864768b1fab645b31491 ./tokenizer_config.json
|
51 |
6b2e0a735969660e720c27061ef3f3d3 ./special_tokens_map.json
|
|
|
52 |
```
|
53 |
|
54 |
-
|
|
|
|
|
55 |
|
56 |
```
|
57 |
python xor_codec.py oasst-rlhf-2-llama-30b-7k-steps/ oasst-rlhf-2-llama-30b-7k-steps-xor/ llama30b_hf/
|
58 |
```
|
59 |
|
60 |
-
You should expect to see one warning message during execution:
|
61 |
|
62 |
`Exception when processing 'added_tokens.json'`
|
63 |
|
64 |
-
This is normal. If similar messages appear for other files, something has gone wrong
|
65 |
|
66 |
-
Now run `find -type f -exec md5sum "{}"
|
67 |
|
68 |
```
|
69 |
d08594778f00abe70b93899628e41246 ./pytorch_model-00007-of-00007.bin
|
@@ -83,4 +142,10 @@ ed59bfee4e87b9193fea5897d610ab24 ./tokenizer_config.json
|
|
83 |
ed991042b2a449123824f689bb94b29e ./pytorch_model-00002-of-00007.bin
|
84 |
```
|
85 |
|
86 |
-
If so you have successfully decoded the weights and should be able to use the model with HuggingFace Transformers.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2 |
license: other
|
3 |
---
|
4 |
|
5 |
+
# OpenAssistant LLaMA 30B RLHF 2
|
6 |
|
7 |
+
Due to the license attached to LLaMA models by Meta AI it is not possible to directly distribute LLaMA-based models. Instead we provide XOR weights for the OA models.
|
8 |
|
9 |
Thanks to Mick for writing the `xor_codec.py` script which enables this process
|
10 |
|
|
|
12 |
|
13 |
Note: This process applies to `oasst-rlhf-2-llama-30b-7k-steps` model. The same process can be applied to other models in future, but the checksums will be different..
|
14 |
|
15 |
+
**This process is tested only on Linux (specifically Ubuntu). Some users have reported that the process does not work on Windows. We recommend using WSL if you only have a Windows machine.**
|
16 |
|
17 |
+
**Some users have also had problems with line endings causing JSON XORs not to work. Please ensure your JSON files have LF (not CR/LF) line endings.**
|
18 |
+
|
19 |
+
To use OpenAssistant LLaMA-Based Models, you should have a copy of the original LLaMA model weights and add them to a `llama` subdirectory here. If you cannot obtain the original LLaMA, see the note in italic below for a possible alternative.
|
20 |
+
|
21 |
+
Ensure your LLaMA 30B checkpoint matches the correct md5sums:
|
22 |
|
23 |
```
|
24 |
f856e9d99c30855d6ead4d00cc3a5573 consolidated.00.pth
|
|
|
28 |
4babdbd05b8923226a9e9622492054b6 params.json
|
29 |
```
|
30 |
|
31 |
+
*If you do not have a copy of the original LLaMA weights and cannot obtain one, you may still be able to complete this process. Some users have reported that [this model](https://huggingface.co/elinas/llama-30b-hf-transformers-4.29) can be used as a base for the XOR conversion. This will also allow you to skip to Step 7. However, we only support conversion starting from LLaMA original checkpoint and cannot provide support if you experience issues with this alternative approach.*
|
32 |
+
|
33 |
+
**Important: Follow these exact steps to convert your original LLaMA checkpoint to a HuggingFace Transformers-compatible format. If you use the wrong versions of any dependency, you risk ending up with weights which are not compatible with the XOR files.**
|
34 |
+
|
35 |
+
1. Create a clean Python **3.10** virtual environment & activate it:
|
36 |
+
|
37 |
+
```
|
38 |
+
python3.10 -m venv xor_venv
|
39 |
+
source xor_venv/bin/activate
|
40 |
+
```
|
41 |
+
|
42 |
+
2. Clone transformers repo and switch to tested version:
|
43 |
+
|
44 |
+
```
|
45 |
+
git clone https://github.com/huggingface/transformers.git
|
46 |
+
cd transformers
|
47 |
+
git checkout d04ec99bec8a0b432fc03ed60cea9a1a20ebaf3c
|
48 |
+
pip install .
|
49 |
+
```
|
50 |
+
|
51 |
+
3. Install **exactly** these dependency versions:
|
52 |
|
53 |
+
```
|
54 |
+
pip install torch==1.13.1 accelerate==0.18.0 sentencepiece==0.1.98 protobuf==3.20.1
|
55 |
+
```
|
56 |
+
|
57 |
+
4. Check `pip freeze` output:
|
58 |
+
|
59 |
+
```
|
60 |
+
accelerate==0.18.0
|
61 |
+
certifi==2022.12.7
|
62 |
+
charset-normalizer==3.1.0
|
63 |
+
filelock==3.12.0
|
64 |
+
huggingface-hub==0.13.4
|
65 |
+
idna==3.4
|
66 |
+
numpy==1.24.2
|
67 |
+
nvidia-cublas-cu11==11.10.3.66
|
68 |
+
nvidia-cuda-nvrtc-cu11==11.7.99
|
69 |
+
nvidia-cuda-runtime-cu11==11.7.99
|
70 |
+
nvidia-cudnn-cu11==8.5.0.96
|
71 |
+
packaging==23.1
|
72 |
+
protobuf==3.20.1
|
73 |
+
psutil==5.9.5
|
74 |
+
PyYAML==6.0
|
75 |
+
regex==2023.3.23
|
76 |
+
requests==2.28.2
|
77 |
+
sentencepiece==0.1.98
|
78 |
+
tokenizers==0.13.3
|
79 |
+
torch==1.13.1
|
80 |
+
tqdm==4.65.0
|
81 |
+
transformers @ file:///mnt/data/koepf/transformers
|
82 |
+
typing_extensions==4.5.0
|
83 |
+
urllib3==1.26.15
|
84 |
+
```
|
85 |
+
|
86 |
+
5. While in `transformers` repo root, run HF LLaMA conversion script:
|
87 |
|
88 |
```
|
89 |
+
python src/transformers/models/llama/convert_llama_weights_to_hf.py --input_dir <input_path_llama_base> --output_dir <output_path_llama30b_hf> --model_size 30B
|
90 |
```
|
91 |
|
92 |
+
6. Run `find . -type f -exec md5sum "{}" +` in the conversion target directory (`output_dir`). This should produce exactly the following checksums if your files are correct:
|
93 |
|
94 |
```
|
95 |
+
462a2d07f65776f27c0facfa2affb9f9 ./pytorch_model-00007-of-00007.bin
|
96 |
+
e1dc8c48a65279fb1fbccff14562e6a3 ./pytorch_model-00003-of-00007.bin
|
97 |
+
9cffb1aeba11b16da84b56abb773d099 ./pytorch_model-00001-of-00007.bin
|
|
|
|
|
|
|
|
|
98 |
aee09e21813368c49baaece120125ae3 ./generation_config.json
|
99 |
+
92754d6c6f291819ffc3dfcaf470f541 ./pytorch_model-00005-of-00007.bin
|
100 |
+
3eddc6fc02c0172d38727e5826181adb ./pytorch_model-00004-of-00007.bin
|
101 |
eeec4125e9c7560836b4873b6f8e3025 ./tokenizer.model
|
102 |
+
99762d59efa6b96599e863893cf2da02 ./pytorch_model-00006-of-00007.bin
|
103 |
598538f18fed1877b41f77de034c0c8a ./config.json
|
104 |
fdb311c39b8659a5d5c1991339bafc09 ./tokenizer.json
|
105 |
+
fecfda4fba7bfd911e187a85db5fa2ef ./pytorch_model.bin.index.json
|
106 |
edd1a5897748864768b1fab645b31491 ./tokenizer_config.json
|
107 |
6b2e0a735969660e720c27061ef3f3d3 ./special_tokens_map.json
|
108 |
+
5cfcb78b908ffa02e681cce69dbe4303 ./pytorch_model-00002-of-00007.bin
|
109 |
```
|
110 |
|
111 |
+
**Important: You should now have the correct LLaMA weights and be ready to apply the XORs. If the checksums above do not match yours, there is a problem.**
|
112 |
+
|
113 |
+
7. Once you have LLaMA weights in the correct format, you can apply the XOR decoding:
|
114 |
|
115 |
```
|
116 |
python xor_codec.py oasst-rlhf-2-llama-30b-7k-steps/ oasst-rlhf-2-llama-30b-7k-steps-xor/ llama30b_hf/
|
117 |
```
|
118 |
|
119 |
+
You should **expect to see one warning message** during execution:
|
120 |
|
121 |
`Exception when processing 'added_tokens.json'`
|
122 |
|
123 |
+
This is normal. **If similar messages appear for other files, something has gone wrong**.
|
124 |
|
125 |
+
8. Now run `find . -type f -exec md5sum "{}" +` in the output directory (here `oasst-rlhf-2-llama-30b-7k-steps`). You should get a file with exactly these checksums:
|
126 |
|
127 |
```
|
128 |
d08594778f00abe70b93899628e41246 ./pytorch_model-00007-of-00007.bin
|
|
|
142 |
ed991042b2a449123824f689bb94b29e ./pytorch_model-00002-of-00007.bin
|
143 |
```
|
144 |
|
145 |
+
If so you have successfully decoded the weights and should be able to use the model with HuggingFace Transformers. **If your checksums do not match those above, there is a problem.**
|
146 |
+
|
147 |
+
- **OASST dataset paper:** https://arxiv.org/abs/2304.07327
|
148 |
+
|
149 |
+
---
|
150 |
+
license: other
|
151 |
+
---
|