-
Notifications
You must be signed in to change notification settings - Fork 11.5k
convert : improve model arch handling #13122
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
base: master
Are you sure you want to change the base?
Conversation
convert_hf_to_gguf.py
Outdated
@@ -1078,8 +1081,12 @@ def __init__(self, *args, **kwargs): | |||
raise TypeError("VisionModel must be subclassed with model_arch = gguf.MODEL_ARCH.CLIP_VISION") | |||
|
|||
# small hack to correct the number of layers | |||
self.tensor_map = gguf.get_tensor_name_map(gguf.MODEL_ARCH.CLIP_VISION, 128) | |||
self.n_embd_text = self.find_hparam(["hidden_size", "n_embd"]) | |||
self.block_count = 512 # vision models are small, this "ought to be enough for anybody" |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
"small" with up to 512 layers :)
What is the max you've seen in the wild?
I think this number could be taken from the config, since vision_config
seems to contain num_hidden_layers
at least for Llama-3.2-11B-Vision. (there's also num_global_layers
, I guess the max of the layer counts should be used)
What model doesn't specify how many layers the vision part has?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I never seen any model having more than 64 layers, but I'm just putting this number here for future proof.
We could get this number from the config, but the problem is that many config.json
nowadays misses that number, as transformers
library omit it if it's the same as default value. For example, this, this and this where you cannot find num_hidden_layers
in vision_config
The frustrating thing is that this start to happen on some text_config
too.
One way to fix this could be to use AutoConfig
, but this won't work on models not transformers
library. While I'm pretty sure this kind of model is rare nowadays, I can't know for sure if people still using it. WDYT?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
One way to fix this could be to use
AutoConfig
,
That could be the way to go, since convert_hf_to_gguf.py
already has hf
in the name and mostly already expects HF-like models which are supported by transformers
. I don't know how much it would change how hparams
is used in set_gguf_parameters
, though.
but this won't work on models not
transformers
library. While I'm pretty sure this kind of model is rare nowadays, I can't know for sure if people still using it. WDYT?
I guess if this is a problem, (e.g. for very new architectures), it could always be possible to temporarily define a PreTrainedConfig
and use AutoConfig.register
.
But! Something which seems important is that AutoConfig
uses the model_type
field instead of archictectures
field, which may change the assumptions in a bunch of places. I'm not sure if it would really be compatible with the idea of using sub-architectures like in this PR.
I guess it's probably fine to keep a high block count, but it makes the tensor map dict bigger than it needs to be.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I don't know how much it would change how
hparams
is used inset_gguf_parameters
, though.
It should not change much, AutoConfig
has a to_dict() function which basically returns the same config.json
, but will all hparams pre-filled.
The simple plan is to replace load_hparams
from open(...)
to AutoConfig.from_pretrained(dir_model).to_dict()
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Here is what I mean: 4840d2f
It does work better now, num_hidden_layers
is no longer missing. However, for smolvlm, some configs are still missing entirely in the to_dict()
, like num_attention_heads
or hidden_size
. Though I think it's not very important for now. Alternative way, we can get them from AutoConfig
object before to_dict()
convert_hf_to_gguf.py
Outdated
hparams["architectures"] = architectures | ||
return hparams | ||
try: | ||
return AutoConfig.from_pretrained(dir_model, trust_remote_code=True).to_dict() |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
return AutoConfig.from_pretrained(dir_model, trust_remote_code=True).to_dict() | |
return AutoConfig.from_pretrained(dir_model, trust_remote_code=False).to_dict() |
I don't think trust_remote_code=True
should be the default here.
If a model uses a custom module, then hopefully it also has the relevant information in config.json
(and/or we can assume some defaults as usual).
I would prefer to avoid running arbitrary code from the config of the models.
(Rethinking about this, and maybe removing that trust would (partially) defeat the purpose of using AutoConfig
...)
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The reason why it left that trust_remote_code=True
was because we already had it in some places in the code, mostly to load the tokenizer. But on second thought, yeah this can be a huge security risk. Basically any bad actors can just trick user to try their model, and they can add a command execution inside.
I think the fallback to config.json
should work well for now, given that not many models need trust_remote_code
One thing I'm a bit concern though, should we guard all other places with trust_remote_code=True
behind a flag, like for example --trust-remote-code
? This is good for performance, but can be quite a bad UX for people don't know why it's needed
This improves the case where
architectures
is defined in bothvision_config
andtext_config
. For example:The arch will be mapped correctly:
ModelType.TEXT
-->Qwen2ForCausalLM
ModelType.VISION
-->InternVisionModel
In case the arch in the sub-config is missing, we simply fallback to the arch in the root-level config. Example:
ModelType.TEXT
-->Qwen2ForCausalLM
ModelType.VISION
-->SmolVLMForConditionalGeneration
This is also the same case where
"architectures": null