SHA256
为了保证文件的完整性，请一定要检查下列文件SHA256值的一致性。
To ensure the completeness of the model, please check the folllowing SHA256 before using them.
Original LLaMA (by Meta AI)
consolidated.*.pth format (original)
下表展示了Facebook发布的原版英文LLaMA的SHA256。
The followings are SHA256 values for the original LLaMA files.
| Model | Parts | SHA256 (consolidated.*.pth) |
|---|---|---|
| 7B | 00 | 700df0d3013b703a806d2ae7f1bfb8e59814e3d06ae78be0c66368a50059f33d |
| 13B | 00 | 745bf4e29a4dd6f411e72976d92b452da1b49168a4f41c951cfcc8051823cf08 |
| 13B | 01 | d5ccbcc465c71c0de439a5aeffebe8344c68a519bce70bc7f9f92654ee567085 |
| 33B | 00 | e23294a58552d8cdec5b7e8abb87993b97ea6eced4178ff2697c02472539d067 |
| 33B | 01 | 4e077b7136c7ae2302e954860cf64930458d3076fcde9443f4d0e939e95903ff |
| 33B | 02 | 24a87f01028cbd3a12de551dcedb712346c0b5cbdeff1454e0ddf2df9b675378 |
| 33B | 03 | 1adfcef71420886119544949767f6a56cb6339b4d5fcde755d80fe68b49de93b |
pytorch.bin format (huggingface)
下表展示了原版英文LLaMA转换为HF格式之后的SHA256。如果你要使用huggingface上的模型，请务必比对。
The followings are SHA256 values for the original LLaMA files (HF format). If you want to use the models on huggingface model hub, make sure to check these values.
| Model | SHA256 (pytorch_model-*.bin) |
|---|---|
| 7B | 0087155d6df07106c1d910bfeb6aab1be8e612dfbf2b56ddfb4ccbde7dbd50d0 461bc5e50200db7813ff99cc0b9316c48ccbd6aaaa31bf8cf7bee0b64bc3eda3 |
| 13B | dd20cdee2637408c6ab88c13f5c27d153bacd0e99f2d55f6a66fbd0269944436 1aba886c5f28d2e2e9b04ab3c4f5bc250c6b06efc1baa3f557677b3097f70e6a 2efc56cddb7877c830bc5b402ee72996aedb5694c9e8007bf1d52d72c0d97d26 |
| 33B | 9c2a7223ab5f9cf3d46913d2b776e99cbd6ed93f69991594b92a8cef0c681a78 4984274738e52195f4b1a5b35d719cf0fade6df1f645507d92d61af4dd8dcdfe 64c73932562810c5c33b15bfec5921d3ced0e8cdb3766c214eda2f45fa3edd13 c7d72d11770f5b58eb45c2dd8e19aae2cbd5a03463b564de3945b21825ebacba 174128542031f4ad7ceb6c799e8e5461ec1ca91a72a01402c567e5f6a8b33d8c 80e2cfa18994385fa88f03d500017346dfd6dc1e58e957d046af39d9a7e254fa 065611608159615ced8d38473ee693129a1a0d872ced0ad8daf09290af7c7061 |
Our LLaMA/Alpaca Model
Tokenizer.model
下表展示了tokenizer.model的SHA256。请注意LLaMA与Alpaca的tokenizer.model不同。对于同一个模型类型，不同大小或者版本的tokenizer.model是相同的。例如，LLaMA-7B, LLaMA-13B, LLaMA-Plus-7B的tokenizer.model相同。
The followings are SHA256 values for tokenizer.model files. Note that tokenizer.model for LLaMA and Alpaca differ. However, different sizes or versions of LLaMA/Alpaca have the same tokenizer.model. For example, LLaMA-7B, LLaMA-13B, LLaMA-Plus-7B's tokenizer.model are the same.
| Model Type | SHA256 |
|---|---|
| LLaMA (7B, 13B, 33B) | e2676d4ca29ca1750f6ff203328d73b189321dc5776ceede037cbd36541d70c0 |
| Alpaca (7B, 13B, 33B) | 2d967e855b1213a439df6c8ce2791f869c84b4f3b6cfacf22b86440b8192a2f8 |
LoRA weight file: adapter_model.bin
下表展示了LoRA主体权重文件adapter_model.bin的SHA256。
The followings are SHA256 values for adapter_model.bin files.
| LoRA Model (adapter_model.bin) | SHA256 |
|---|---|
| Chinese-LLaMA-7B | 2a2c24d096f5d509f24946fdbd8c25e1ce4a0acb955902f7436d74c0c0379d86 |
| Chinese-LLaMA-Plus-7B | 8c928db86b2a0cf73f019832f921eb7e1e069ca21441b4bfa12c4381c6cc46be |
| Chinese-LLaMA-13B | 6a4ce789d219bde122f8d9a20371937f2aa2ee86a2311d9f5e303df2e774f9fc |
| Chinese-LLaMA-Plus-13B | 784fcff9c4bdf4e77d442a01158e121caf8fcce0f97ffb32396fe7a3617ee7e8 |
| Chinese-LLaMA-33B | 93a449bafb71ff1bb74a4a21e64e102e5078e5c3898eb40d013790072a0fa3de |
| Chinese-Alpaca-7B | 0d9b6ed8e4a7d1ae590a16c89a452a488d66ff07e45487972f61c2b6e46e36de |
| Chinese-Alpaca-Plus-7B | 4ee0bf805c312a9a771624d481fbdb4485e1b0a70cd2a8da9f96137f177b795d |
| Chinese-Alpaca-13B | cb8dda3c005f3343a0740dcd7237fbb600cb14b6bff9b6f3d488c086a2f08ada |
| Chinese-Alpaca-Plus-13B | a1fcdcb6d7e1068f925fb36ec78632c76058ba12ba352bed4d44060b8e6f4706 |
| Chinese-Alpaca-33B | 6b39da4c682e715a9de30b247b7e9b812d2d54f7d320ec9b452000a5cd4d178d |
Merged files (consolidated.*.pth)
下表展示了合并LoRA权重后的全量模型权重（PyTorch版）的SHA256。PyTorch版本不影响实际权重数据，但影响meta信息，所以SHA256也会不同。建议合并模型时使用PyTorch >= 1.13.0版本，以确保以下SHA256有参考性。
⚠️ 请优先确保合并前的基模型和LoRA权重的SHA256是否与上述表中所述值一致。
The followings are SHA256 values for merged files (consolidated.*.pth). Note that the version of PyTorch does not affect actual weights but meta informations are slightly different. Please check SHA256 according to PyTorch version >= 1.13.0.
| Model | SHA256 (PyTorch >= 1.13.0) |
|---|---|
| Chinese-LLaMA-7B | 245427a306e3253db3f534e2a1d7548a8eb781ae8761f9e98979b4aced6b43d8 |
| Chinese-LLaMA-Plus-7B | f8d380d63f77a08b7f447f5ec63f0bb1cde9ddeae2207e9f86e6b5f0f95a7955 |
| Chinese-LLaMA-13B | aa7f4599487ea2b0d0aca2b522c39370897f9afd9839aac7d02155957f1f019f 3954f3e7f7264994f23800a04423e6563cc1959ac699d9eaaa6801b4f9392ebd |
| Chinese-LLaMA-Plus-13B | 4de7d188003c778f216342de2dc5c9a9c74278c701c63a7b6bcd7957f5ebfdf5 ff8046f9eb8b05dd86597c21edd07894aec00b31842a4c11996a4003091ea7c9 |
| Chinese-LLaMA-33B | 054e9b7dffa3b92a053ca32acac6e22b27c184ed2b8563f8e44e6570ba416357 a0fe86c45a0819f45a509776d82778b7de75fbff8d37afa97159b24de5448b7b 13df5f74dc7bc1204076b1febef818fb3cec978de27bf8fc85c70e7d62282df9 f4f28106c343c5804613faa9852f29fbc60764366bcb0d37ef2811a17be2d336 |
| Chinese-Alpaca-7B | fbfccc91183169842aac8d093379f0a449b5a26c5ee7a298baf0d556f1499b90 |
| Chinese-Alpaca-Plus-7B | 8b8f6551d0d83f93e378622b9f8dad0bec189da6c29d8a78de493e6aee9bd35f |
| Chinese-Alpaca-13B | 30cefb5be9091c3e17fbba5d91bf16266a2ddf86cde53370a9982b232ff8a2f4 ce946742b0f122f472e192c3f77d506e0c26578b4b881d07d919553333affecd |
| Chinese-Alpaca-Plus-13B | 1834558214c1dddc0d8b2826ece086908b9d2293241d0e12cecb48a035ec561b bf70001600ce166f6ca4ef59df5510f0582cdc119fb74e27d9cf3e4c7b142015 |
| Chinese-Alpaca-33B | 72bfe67481c0df1b8c3b536acd15ac42c1163b0727b1beb6409ee31d14cb2490 fd2151ea714a6e0706a60cca5ab7abf8558e665d4cb001481c6df616c0821c16 4a7e3de6881769f9c2413f0867e67da20efdf4502602ab90483cb99c593e51ed 99c81a7a310802dcc579fe96288fbc18d4486f92020eaf925e1c33db8311378a |
How To Check SHA256
In MacOS,
> shasum -a 256 your-model-file
In Linux,
> sha256sum your-model-file
In Windows,
> certutil -hashfile your-model-file sha256