SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity

Add InternVL (2.5 MPO) (#35968)

Browse Source 
* initial commit

* add convert internvl

* add first end-to-end working internvl

* nit prompt and image proc

* add working chat template

* add conversion llama-based models

* add tests

* pass all tests

* fix isort

* fix modular after main merge

* add video processing for internvl

* add support for interlaced images and videos

* Remove processing and config from modular, add more tests

* add llama model tests

* Modify processor for compatibility with refactored got ocr image processor

* add comments in processor

* Add docs and nits

* change video processing to use custom sample_indices_fn

* rebase and fix tests

* add processor tests

* Add changes Raushan review

* Use the new attention interface for the vision model

* nits

* add support for custom video_load_backend

* remove mention to InternVLTokenizer

* refactor vision model to simplify logic

* refactor processor for better readibility

* fix copies

* fix require av processor test

* refactor internVL vision

* Update processor and fix processing tests

* fix docstring

* update convert_weights for internvl3

* change image processor to fast by default

* remove do_center_crop=True in convert_weights

* force use_cache to True

* push_to_hub before reloading

* fix internVLVision for larger models

* update convert weight for qk norm

* fix convert_weights

* fix eos_token_id in convert

* update docs and integration tests

* make modifs after review

* fix wrong k_norm and reduce modular

* change image_token_index to image_token_id

* change checkpoint to OpenGVLab org

* last nits

* explicitely del self.num_key_value_groups

* add extra special tokens
This commit is contained in:
Yoni Gozlan
2025-04-18 18:57:33 +02:00
committed by GitHub
parent b0c6ff5e13
commit a245011252
20 changed files with 4447 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
docs/source/en/perf_infer_gpu_one.md
View File

@@ -244,7 +244,7 @@ model = AutoModelForCausalLM.from_pretrained("meta-llama/Llama-3.1-8B", device_m
### Benchmarks
FlashAttention2 speeds up inference considerably especially for inputs with long sequences. However, since FlashAttention2 doesn't support computing attention scores with padding tokens, you must manually pad and unpad the attention scores for batched inference if a sequence contains padding tokens. The downside is batched generation is slower with padding tokens.
FlashAttention2 speeds up inference considerably especially for inputs with long sequences. However, since FlashAttention2 doesn't support computing attention scores with padding tokens, you must manually pad and unpad the attention scores for batched inference if a sequence contains padding tokens. The downside is batched generation is slower with padding tokens.
<hfoptions id="padded">
<hfoption id="short sequence length">