model type `deepseek_v32` not found
from transformers import AutoTokenizer, AutoModelForCausalLM, AutoConfig
model_name = "deepseek-ai/DeepSeek-V3.2"
config = AutoConfig.from_pretrained(model_name, trust_remote_code=True)
KeyError: 'deepseek_v32'
During handling of the above exception, another exception occurred:
ValueError Traceback (most recent call last)
Cell In[3], line 1
----> 1 config = AutoConfig.from_pretrained(model_name, trust_remote_code=True)
File ~/miniconda3/envs/agentic/lib/python3.12/site-packages/transformers/models/auto/configuration_auto.py:1362, in AutoConfig.from_pretrained(cls, pretrained_model_name_or_path, **kwargs)
1360 config_class = CONFIG_MAPPING[config_dict["model_type"]]
1361 except KeyError:
-> 1362 raise ValueError(
1363 f"The checkpoint you are trying to load has model type `{config_dict['model_type']}` "
1364 "but Transformers does not recognize this architecture. This could be because of an "
1365 "issue with the checkpoint, or because your version of Transformers is out of date.\n\n"
1366 "You can update Transformers with the command `pip install --upgrade transformers`. If this "
1367 "does not work, and the checkpoint is very new, then there may not be a release version "
1368 "that supports this model yet. In this case, you can get the most up-to-date code by installing "
1369 "Transformers from source with the command "
1370 "`pip install git+https://github.com/huggingface/transformers.git`"
1371 )
1372 return config_class.from_dict(config_dict, **unused_kwargs)
1373 else:
1374 # Fallback: use pattern matching on the string.
1375 # We go from longer names to shorter names to catch roberta before bert (for instance)
ValueError: The checkpoint you are trying to load has model type `deepseek_v32` 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 This could be because of an issue with the checkpoint, or because your version of Transformers is out of date.
You can update Transformers with the command `pip install --upgrade transformers`. If this does not work, and the checkpoint is very new, then there may not be a release version that supports this model yet. In this case, you can get the most up-to-date code by installing Transformers from source with the command `pip install git+https://github.com/huggingface/transformers.git`
I cannot use deepseek_v32 model type. How to use this?
As the error stated, your transformers package did not support the architecture yet. Upgrade your package to resolve.
ValueError: The checkpoint you are trying to load has model type deepseek_v32 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.
You can update Transformers with the command pip install --upgrade transformers. If this does not work, and the checkpoint is very new, then there may not be a release version that supports this model yet. In this case, you can get the most up-to-date code by installing Transformers from source with the command pip install git+https://github.com/huggingface/transformers.git
Ignore my previous message, please take a look at the folder inference you should have checked out from hf.
see the following url
https://huggingface.co/deepseek-ai/DeepSeek-V3.2/tree/main/inference
did you install the inference package it shipped with?
I assume you do speak Chinese, so here is that
宝啊，和模型一起的推理包你安装了吗？要用和模型一起的那个推理包，你应该是有下载了的。
@lianxuI tried to quantize the weights with the Intel Autoround library which depends on Transformers and got the same error that deepseek-v32 is not supported. Even on the latest nighlty build of Transformers built from source
@lianxuI tried to quantize the weights with the Intel Autoround library which depends on Transformers and got the same error that deepseek-v32 is not supported. Even on the latest nighlty build of Transformers built from source
Are you using the inference package as I mentioned above? There is a Transformer implementation inside the inference package this model shipped with. Take a look inside its source code please. The transformers package you installed via pip DID NOT ADOPT this new deepseek model. You have to use what it provided within this model's repo.