Skip to content

[examples] fix vision_tower/multi_modal_projector regexes #1871

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.

Sign up for GitHub

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

Merged
merged 13 commits into from
Oct 1, 2025
Merged

[examples] fix vision_tower/multi_modal_projector regexes #1871

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
13 commits
Select commit Hold shift + click to select a range
49fca17
fix vision_tower/multi_modal_projector regexes
brian-dellabetta Sep 26, 2025
8cdef59
vision_model as well
brian-dellabetta Sep 29, 2025
4aab7ed
Merge branch 'main' into bdellabe/examples-fix-ignores
brian-dellabetta Sep 29, 2025
6127a0d
Merge branch 'main' into bdellabe/examples-fix-ignores
brian-dellabetta Sep 30, 2025
e1f32d5
Update examples/quantization_w4a4_fp4/llama4_example.py
brian-dellabetta Sep 30, 2025
273d34e
Update tests/e2e/vLLM/recipes/FP8/recipe_fp8_dynamic.yaml
brian-dellabetta Sep 30, 2025
845176f
Apply suggestion from @fynnsu
brian-dellabetta Sep 30, 2025
d3321e9
Apply suggestion from @fynnsu
brian-dellabetta Sep 30, 2025
713c9eb
Apply suggestion from @fynnsu
brian-dellabetta Sep 30, 2025
fcbc2f3
Apply suggestion from @fynnsu
brian-dellabetta Sep 30, 2025
cd0ea9e
Merge branch 'main' into bdellabe/examples-fix-ignores
brian-dellabetta Sep 30, 2025
26cd06e
Merge branch 'main' into bdellabe/examples-fix-ignores
brian-dellabetta Oct 1, 2025
bb55ca0
Merge branch 'main' into bdellabe/examples-fix-ignores
brian-dellabetta Oct 1, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 1 addition & 1 deletion examples/multimodal_vision/README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -37,7 +37,7 @@ recipe = [
targets="Linear",
scheme="W4A16",
sequential_targets=["MistralDecoderLayer"],
ignore=["re:.*lm_head", "re:vision_tower.*", "re:multi_modal_projector.*"],
ignore=["re:.*lm_head", "re:.*vision_tower.*", "re:.*multi_modal_projector.*"],
),
]
```
Expand Down
12 changes: 7 additions & 5 deletions examples/multimodal_vision/llama4_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,9 +52,11 @@ def preprocess_function(example):
def data_collator(batch):
assert len(batch) == 1
return {
key: torch.tensor(value)
if key != "pixel_values"
else torch.tensor(value, dtype=torch.bfloat16).squeeze(0)
key: (
torch.tensor(value)
if key != "pixel_values"
else torch.tensor(value, dtype=torch.bfloat16).squeeze(0)
)
for key, value in batch[0].items()
}

Expand All @@ -67,8 +69,8 @@ def data_collator(batch):
"re:.*lm_head",
"re:.*self_attn",
"re:.*router",
"re:vision_model.*",
"re:multi_modal_projector.*",
"re:.*vision_model.*",
"re:.*multi_modal_projector.*",
"Llama4TextAttention",
],
)
Expand Down
2 changes: 1 addition & 1 deletion examples/multimodal_vision/llava_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,7 @@ def data_collator(batch):
GPTQModifier(
targets="Linear",
scheme="W4A16",
ignore=["re:.*lm_head", "re:vision_tower.*", "re:multi_modal_projector.*"],
ignore=["re:.*lm_head", "re:.*vision_tower.*", "re:.*multi_modal_projector.*"],
),
]

Expand Down
10 changes: 6 additions & 4 deletions examples/multimodal_vision/mistral3_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,9 +31,11 @@
def data_collator(batch):
assert len(batch) == 1
return {
key: torch.tensor(value)
if key != "pixel_values"
else torch.tensor(value, dtype=model.dtype)
key: (
torch.tensor(value)
if key != "pixel_values"
else torch.tensor(value, dtype=model.dtype)
)
for key, value in batch[0].items()
}

Expand All @@ -43,7 +45,7 @@ def data_collator(batch):
GPTQModifier(
targets="Linear",
scheme="W4A16",
ignore=["re:.*lm_head", "re:vision_tower.*", "re:multi_modal_projector.*"],
ignore=["re:.*lm_head", "re:.*vision_tower.*", "re:.*multi_modal_projector.*"],
),
]

Expand Down
2 changes: 1 addition & 1 deletion examples/multimodal_vision/mllama_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,7 @@ def data_collator(batch):
GPTQModifier(
targets="Linear",
scheme="W4A16",
ignore=["re:.*lm_head", "re:multi_modal_projector.*", "re:vision_model.*"],
ignore=["re:.*lm_head", "re:.*multi_modal_projector.*", "re:.*vision_model.*"],
),
]

Expand Down
2 changes: 1 addition & 1 deletion examples/multimodal_vision/pixtral_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,7 +36,7 @@ def data_collator(batch):
GPTQModifier(
targets="Linear",
scheme="W4A16",
ignore=["re:.*lm_head", "re:vision_tower.*", "re:multi_modal_projector.*"],
ignore=["re:.*lm_head", "re:.*vision_tower.*", "re:.*multi_modal_projector.*"],
),
]

Expand Down
12 changes: 7 additions & 5 deletions examples/quantization_w4a4_fp4/llama4_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -52,9 +52,11 @@ def preprocess_function(example):
def data_collator(batch):
assert len(batch) == 1
return {
key: torch.tensor(value)
if key != "pixel_values"
else torch.tensor(value, dtype=torch.bfloat16).squeeze(0)
key: (
torch.tensor(value)
if key != "pixel_values"
else torch.tensor(value, dtype=torch.bfloat16).squeeze(0)
)
for key, value in batch[0].items()
}

Expand All @@ -67,8 +69,8 @@ def data_collator(batch):
"re:.*lm_head",
"re:.*self_attn",
"re:.*router",
"re:vision_model.*",
"re:multi_modal_projector.*",
"re:.*vision_model.*",
"re:.*multi_modal_projector.*",
"Llama4TextAttention",
],
)
Expand Down
2 changes: 1 addition & 1 deletion examples/quantization_w8a8_fp8/llama3.2_vision_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@
recipe = QuantizationModifier(
targets="Linear",
scheme="FP8_DYNAMIC",
ignore=["re:.*lm_head", "re:multi_modal_projector.*", "re:vision_model.*"],
ignore=["re:.*lm_head", "re:.*multi_modal_projector.*", "re:.*vision_model.*"],
)

# Apply quantization and save to disk in compressed-tensors format.
Expand Down
4 changes: 2 additions & 2 deletions examples/quantization_w8a8_fp8/llama4_fp8_block_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -22,8 +22,8 @@
"re:.*lm_head",
"re:.*self_attn",
"re:.*router",
"re:vision_model.*",
"re:multi_modal_projector.*",
"re:.*vision_model.*",
"re:.*multi_modal_projector.*",
"Llama4TextAttention",
],
)
Expand Down
2 changes: 1 addition & 1 deletion examples/quantization_w8a8_fp8/llava1.5_example.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -17,7 +17,7 @@
recipe = QuantizationModifier(
targets="Linear",
scheme="FP8_DYNAMIC",
ignore=["re:.*lm_head", "re:multi_modal_projector.*", "re:vision_tower.*"],
ignore=["re:.*lm_head", "re:.*multi_modal_projector.*", "re:.*vision_tower.*"],
)

# Apply quantization and save to disk in compressed-tensors format.
Expand Down
2 changes: 1 addition & 1 deletion tests/e2e/vLLM/recipes/FP8/recipe_fp8_dynamic.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
quant_stage:
quant_modifiers:
QuantizationModifier:
ignore: ["lm_head", "re:vision_tower.*", "re:multi_modal_projector.*", "re:visual.*", "re:vision_model.*", "re:model.visual.*"]
ignore: ["lm_head", "re:vision_tower.*", "re:.*multi_modal_projector.*", "re:.*visual.*", "re:.*vision_model.*"]
config_groups:
group_0:
weights:
Expand Down
2 changes: 1 addition & 1 deletion tests/e2e/vLLM/recipes/INT8/recipe_int8_channel_weight_dynamic_per_token.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -3,7 +3,7 @@ quant_stage:
SmoothQuantModifier:
smoothing_strength: 0.8
GPTQModifier:
ignore: ["lm_head", "re:vision_tower.*", "re:multi_modal_projector.*", "re:visual.*", "re:vision_model.*", "re:model.visual.*"]
ignore: ["lm_head", "re:.*vision_tower.*", "re:.*multi_modal_projector.*", "re:.*visual.*", "re:.*vision_model.*"]
actorder: null
config_groups:
group_0:
Expand Down
2 changes: 1 addition & 1 deletion tests/e2e/vLLM/recipes/actorder/recipe_w4a16_actorder_weight.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,7 +1,7 @@
quant_stage:
quant_modifiers:
GPTQModifier:
ignore: ["lm_head", "re:vision_tower.*", "re:multi_modal_projector.*", "re:visual.*", "re:vision_model.*", "re:model.visual.*"]
ignore: ["lm_head", "re:.*vision_tower.*", "re:.*multi_modal_projector.*", "re:.*visual.*", "re:.*vision_model.*"]
actorder: "weight"
config_groups:
group_0:
Expand Down