关于transformers无法识别模型类型 ValueError: The checkpoint you are trying to load has model type `tinyllm` but Transformers does not recognize this architecture.

[1190201205]

在网络上尝试修改transformers版本号 同时修改模型中的config中为对应版本号也没有解决
{
"architectures": [
"TinyllmForCausalLM"
],
"attention_dropout": 0.0,
"hidden_act": "silu",
"hidden_size": 512,
"initializer_range": 0.02,
"intermediate_size": 1408,
"max_position_embeddings": 1024,
"model_type": "tinyllm",
"num_attention_heads": 8,
"num_hidden_layers": 8,
"num_key_value_heads": 8,
"rms_norm_eps": 1e-06,
"rope_theta": 10000.0,
"tie_word_embeddings": false,
"torch_dtype": "float16",
"transformers_version": "4.42.3",
"use_cache": true,
"vocab_size": 64798
}

[rank0]: Traceback (most recent call last):
[rank0]: File "sft_train.py", line 193, in
[rank0]: main()
[rank0]: File "sft_train.py", line 149, in main
[rank0]: config = transformers.AutoConfig.from_pretrained(
[rank0]: File "/home/daichenrui2404/miniconda3/envs/TINY_LLM_ZH/lib/python3.8/site-packages/transformers/models/auto/configuration_auto.py", line 984, in from_pretrained
[rank0]: raise ValueError(
[rank0]: ValueError: The checkpoint you are trying to load has model type tinyllm but Transformers does not recognize this architecture. This could be because of an issue with the checkpoint, or because your version of Transformers is out of date.
E0708 15:05:53.145956 139943936660672 torch/distributed/elastic/multiprocessing/api.py:826] failed (exitcode: 1) local_rank: 0 (pid: 1131943) of binary: /home/daichenrui2404/miniconda3/envs/TINY_LLM_ZH/bin/python
Traceback (most recent call last):
File "/home/daichenrui2404/miniconda3/envs/TINY_LLM_ZH/bin/torchrun", line 8, in
sys.exit(main())
File "/home/daichenrui2404/miniconda3/envs/TINY_LLM_ZH/lib/python3.8/site-packages/torch/distributed/elastic/multiprocessing/errors/init.py", line 347, in wrapper
return f(*args, **kwargs)
File "/home/daichenrui2404/miniconda3/envs/TINY_LLM_ZH/lib/python3.8/site-packages/torch/distributed/run.py", line 879, in main
run(args)
File "/home/daichenrui2404/miniconda3/envs/TINY_LLM_ZH/lib/python3.8/site-packages/torch/distributed/run.py", line 870, in run
elastic_launch(
File "/home/daichenrui2404/miniconda3/envs/TINY_LLM_ZH/lib/python3.8/site-packages/torch/distributed/launcher/api.py", line 132, in call
return launch_agent(self._config, self._entrypoint, list(args))
File "/home/daichenrui2404/miniconda3/envs/TINY_LLM_ZH/lib/python3.8/site-packages/torch/distributed/launcher/api.py", line 263, in launch_agent
raise ChildFailedError(
torch.distributed.elastic.multiprocessing.errors.ChildFailedError:

sft_train.py FAILED

[wdndev]

由于这个模型是我自己到处抄出来的，在transformers官方的代码库中不存在这个模型文件，如果使用 Auto 类 （AutoModelForCausalLM， AutoTokenizer）加载模型，需要指定模型结构类，如下图配置文件所示：

[happy-xlf]

制定之后，在sft训练的时候会报：ValueError: Unrecognized configuration class <class 'transformers_modules.checkpoint-1300.configuration_tinyllm.TinyllmConfig'> to build an AutoTokenizer.
Model type should be one of AlbertConfig, AlignConfig, BarkConfig, BartConfig, BertConfig, BertGenerationConfig, BigBirdConfig, BigBirdPegasusConfig, BioGptConfig, BlenderbotConfig, BlenderbotSmallConfig, BlipConfig, Blip2Config, BloomConfig, BridgeTowerConfig, BrosConfig, CamembertConfig, CanineConfig, ChineseCLIPConfig, ClapConfig, CLIPConfig, CLIPSegConfig, ClvpConfig, LlamaConfig, CodeGenConfig, CohereConfig, ConvBertConfig, CpmAntConfig, CTRLConfig, Data2VecAudioConfig, Data2VecTextConfig, DbrxConfig, DebertaConfig, DebertaV2Config, DistilBertConfig, DPRConfig, ElectraConfig, ErnieConfig, ErnieMConfig, EsmConfig, FalconConfig, FastSpeech2ConformerConfig, FlaubertConfig, FNetConfig, FSMTConfig, FunnelConfig, GemmaConfig, GitConfig, GPT2Config, GPT2Config, GPTBigCodeConfig, GPTNeoConfig, GPTNeoXConfig, GPTNeoXJapaneseConfig, GPTJConfig, GPTSanJapaneseConfig, GroundingDinoConfig, GroupViTConfig, HubertConfig, IBertConfig, IdeficsConfig, Idefics2Config, InstructBlipConfig, JambaConfig, JukeboxConfig, Kosmos2Config, LayoutLMConfig, LayoutLMv2Config, LayoutLMv3Config, LEDC an AutoTokenizer.