Skip to content

Can't falcon_convert on OpenBuddy Falcon 7B model, KeyError [fixed] #68

Description

@tak2hu

Model: OpenBuddy Falcon 7B
python falcon_convert.py openbuddy-falcon-7b-v6-bf16 openbuddy-ggllm use-f32

Error:

* Loading model from:  openbuddy-falcon-7b-v6-bf16
Vocab size: 70144
Hidden size: 4544
Number of heads: 71
Number of layers: 32
Number of head_kv: 1
Number of head_dim: 64
Traceback (most recent call last):
  File "/home/paloma/Git/ggllm.cpp/falcon_convert.py", line 111, in <module>
    text = bytearray([byte_decoder[c] for c in reverse_vocab[i]])
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/home/paloma/Git/ggllm.cpp/falcon_convert.py", line 111, in <listcomp>
    text = bytearray([byte_decoder[c] for c in reverse_vocab[i]])
                      ~~~~~~~~~~~~^^^
KeyError: '能'
s -lh openbuddy-falcon-7b-v6-bf16/
total 13G
-rw-r--r-- 1 paloma paloma  992 Jul 18 19:11 config.json
-rw-r--r-- 1 paloma paloma 2,6K Jul 18 19:11 configuration_RW.py
-rw-r--r-- 1 paloma paloma  111 Jul 18 19:11 generation_config.json
-rw-r--r-- 1 paloma paloma  47K Jul 18 19:11 modelling_RW.py
-rw-r--r-- 1 paloma paloma 9,4G Jul 18 19:05 pytorch_model-00001-of-00002.bin
-rw-r--r-- 1 paloma paloma 3,7G Jul 18 19:05 pytorch_model-00002-of-00002.bin
-rw-r--r-- 1 paloma paloma  17K Jul 18 19:11 pytorch_model.bin.index.json
-rw-r--r-- 1 paloma paloma   28 Jul 18 19:11 README.md
-rw-r--r-- 1 paloma paloma  281 Jul 18 19:11 special_tokens_map.json
-rw-r--r-- 1 paloma paloma  180 Jul 18 19:11 tokenizer_config.json
-rw-r--r-- 1 paloma paloma 3,5M Jul 18 19:11 tokenizer.json

Operating System: Arch Linux

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions