patch 1 for mllm #298
Merged
patch 1 for mllm #298
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Signed-off-by: n1ck-guo <[email protected]>
wenhuach21
reviewed
Nov 4, 2024
…ages Signed-off-by: n1ck-guo <[email protected]>
wenhuach21
reviewed
Nov 5, 2024
wenhuach21
approved these changes
Nov 5, 2024
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
wenhuach21
reviewed
Nov 7, 2024
wenhuach21
reviewed
Nov 7, 2024
Signed-off-by: n1ck-guo <[email protected]>
wenhuach21
reviewed
Nov 7, 2024
wenhuach21
reviewed
Nov 7, 2024
auto_round/script/mllm.py
Outdated
| self.add_argument("--dataset", type=str, default=None, | ||
| help="the dataset for quantization training. It can be a custom one.") | ||
| self.add_argument("--dataset", type=str, default="llava_v1_5_mix665k", | ||
| help="The dataset for quantization training. It can be a custom one.") |
There was a problem hiding this comment.
use lowercase for the first letter to follow our current style
wenhuach21
reviewed
Nov 7, 2024
| processor.chat_template = None | ||
| safe_serialization = True | ||
| if "phi3_v" in model_type: | ||
| safe_serialization = False |
There was a problem hiding this comment.
the code is tricky, better move to model config later
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
for more information, see https://pre-commit.ci
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
wenhuach21
reviewed
Nov 11, 2024
Signed-off-by: n1ck-guo <[email protected]>
wenhuach21
reviewed
Nov 11, 2024
| device=device_str, seed=args.seed, gradient_accumulate_steps=args.gradient_accumulate_steps, | ||
| scale_dtype=args.scale_dtype, layer_config=layer_config, | ||
| scale_dtype=args.scale_dtype, layer_config=layer_config, template=args.template, |
There was a problem hiding this comment.
rename template to prompt_template to make it more easier to understand?
There was a problem hiding this comment.
This template is more like "model series template". It include processor(text/image), data collator, and special token.
wenhuach21
reviewed
Nov 11, 2024
wenhuach21
reviewed
Nov 11, 2024
wenhuach21
reviewed
Nov 11, 2024
wenhuach21
reviewed
Nov 11, 2024
test/test_basic_usage.py
Outdated
| res = os.system( | ||
| f"{python_path} ../auto_round/__main__.py --model 'facebook/opt-125m' --iter 2 --nsamples 1 --format auto_gptq,auto_round --disable_eval --output_dir ./saved") | ||
| f"cd .. && {python_path} -m auto_round --mllm --iter 2 --nsamples 10 --format auto_round --output_dir ./saved") | ||
| if res > 0 or res == -1: | ||
| assert False, "cmd line test fail, please have acheck" |
There was a problem hiding this comment.
add another test for auto-round -h
wenhuach21
reviewed
Nov 11, 2024
auto_round/mllm/mllm_dataset.py
Outdated
| @@ -44,18 +46,26 @@ def register(dataset): | |||
| return register | |||
|
|
|||
|
|
|||
|
|
|||
| @register_dataset("llava") | |||
There was a problem hiding this comment.
better add more information like liuhaotian/llava and 58k or 150k
Signed-off-by: n1ck-guo <[email protected]>
wenhuach21
reviewed
Nov 11, 2024
auto_round/script/mllm.py
Outdated
| @@ -50,8 +50,10 @@ def __init__(self, *args, **kwargs): | |||
| self.add_argument("--asym", action='store_true', | |||
| help="whether to use asym quantization") | |||
|
|
|||
| self.add_argument("--dataset", type=str, default=None, | |||
| help="the dataset for quantization training. It can be a custom one.") | |||
| self.add_argument("--dataset", type=str, default="llava_conv_58k", | |||
There was a problem hiding this comment.
llava_conv_58k ==>better change to liuhaotian/llava_conv_58k
Signed-off-by: n1ck-guo <[email protected]>
Signed-off-by: n1ck-guo <[email protected]>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
TODO: