Add llama4 (#37307)
* remove one of the last deps * update fast image processor after refactor * styling * more quality of life improvements * nit * update * cleanups * some cleanups * vllm updates * update fake image token * [convert] Fix typo * [convert] Strip extraneous bytes from shards * [convert] Minor fixes * [convert] Use num_experts * multi-image fixes in modeling + processor * fixup size * 128 experts * Use default rope * Unfuse mlp * simplify a lot inputs embeds merging * remove .item() 👀 * fix from review * Address feedback * Use None "default" for rope_scaling. Add eot. * set seed * return aspect ratios and bug fixes * Moe 128 rebased (#8) * 128 experts * Use default rope * Unfuse mlp * Address feedback * Use None "default" for rope_scaling. Add eot. * Meta/llama quant compat (#7) * add quant compatible model & conversion code for llama4 * fix a few issues * fix a few issues * minor type mapping fix --------- Co-authored-by: Lu Fang <fanglu@fb.com> * use a new config parameter to determine which model definition to use for MoE --------- Co-authored-by: Pedro Cuenca <pedro@huggingface.co> Co-authored-by: Lu Fang <fanglu@fb.com> * un-comment write_tokenizer from converting script * remove un-used imports * [llama4] Pop aspect_ratios from image processor output in Llama4Processor Signed-off-by: Jon Swenson <jmswen@gmail.com> * Fix parameter_count name * Update src/transformers/models/llama4/configuration_llama4.py * nit * Add changes for no_rope, moe_layers, chunked attention. Just need to test all * Update src/transformers/models/llama4/image_processing_llama4_fast.py * nit * fix post merge with main * support flex attention * fixes * fix * add layer * small updates * rebase and delete llm_compressor * nit * [llama4/mm] Add back <|image|> token that delimits global tile * [llama4/mm] Fix Llama 4 image processing unit tests * add explicit dtype Signed-off-by: Jon Swenson <jmswen@gmail.com> * sdpa works * comment todo small * fix model loading Signed-off-by: Zijing Liu <liuzijing2014@gmail.com> * revert * nits * small fix for TP on 1 node * Read new params from config * Add <|eom|> * lol don't know how this got here * adding fp8 * Save processor, fix chat template * style * Add boi/eoi tokens We don't use them. * fixes for now flex seems to work :) * updates * nits * updates * missking keys * add context parallel * update * update * fix * nits * add worldsize and make eager attn work for vision * Ignore new key present in base models * add tp_plan * fix nope Signed-off-by: Zijing Liu <liuzijing2014@gmail.com> * minor fix Signed-off-by: Zijing Liu <liuzijing2014@gmail.com> * Clean up Llama4 vision model * current updates * add support for `attn_temperature_tuning` * add floor scale * add missing attn scales * push what works, dirty trick for the device synch * oups * Fix pad_token_id See https://huggingface.co/ll-re/Llama-4-Scout-17B-16E/discussions/2/files Confirmed in the original codebase. * fix causallml loading * rm * fix tied-weights * fix sdpa * push current version * should work with both short and long * add compressed_tensos & fix fbgemm tp * Fix flex impl * style * chunking * try to revert the potentially breaking change * fix auto factory * fix shapes in general * rm processing * commit cache utils cleanup * Fix context length * fix * allocate * update tp_plan * fix SDPA! * Add support for sparse `Llama4TextMoe` layer from the kernel hub * cleanup * better merge * update * still broken fixing now * nits * revert print * Write max_position_embeddings and max_model_length * Update modeling_llama4.py * Save attention_chunk_size * Sync eos terminators * Read initializer_range * style * remove `dict` * fix * eager should use `chunked_attention_mask` * revert * fixup * fix config * Revert "Merge pull request #36 from huggingface/sparse-llama4-moe" This reverts commit ccda19f050867dd42ea143c5de60f3dec81375f0, reversing changes made to a515579aed8c0fe9bf529b6c40446a289406d5d6. * Fix typo and remove warning with compiled flex and chunked prefill * Fix MoE vs FF (#41) * fix * Use correct no_rope_layers if provided one is empty list * update tests * fix * skipping some tests * fix fp8 loading Signed-off-by: Zijing Liu <liuzijing2014@gmail.com> * fix text geneartion pipeline Signed-off-by: Zijing Liu <liuzijing2014@gmail.com> * eager needs 4D mask * fix * Some cleanup * fix * update * fix * replace correctly module * patch * modulelist * update * update * clean up * Don't move to `cuda:0` in distributed mode * restrict to compressed tensors for now * rm print * Docs! * Fixes * Update docs/source/en/model_doc/llama4.md Co-authored-by: Pedro Cuenca <pedro@huggingface.co> * Fixes * cuda graph fix * revert some stuff * fixup * styling * Update src/transformers/models/llama4/modeling_llama4.py Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com> * fixup * commit licence, cleanup here and there and style * more styling changes * fix dummies * fix and clean docstrings * remove comment * remove warning * Only fast image processor is supported * nit * trigger CI * fix issue with flex encoder * fix dynamic cache * Code quality * Code quality * fix more tests for now * Code quality * Code quality * Nuke bunch of failing stuff * Code quality * Code quality * cleanup removal of slow image processor * ruff fix fast image processor * fix * fix styling * Docs * Repo consistency * Repo consistency * fix sliding window issue * separate llama cache * styling * Repo consistency * Repo consistency * push waht works * L4 Repo consistency * Docs * fix last last alst alst alst alstsaltlsltlaslt --------- Signed-off-by: Jon Swenson <jmswen@gmail.com> Signed-off-by: Zijing Liu <liuzijing2014@gmail.com> Co-authored-by: yonigozlan <yoni.gozlan10@gmail.com> Co-authored-by: Pedro Cuenca <pedro@huggingface.co> Co-authored-by: Pablo Montalvo <pablo.montalvo.leroux@gmail.com> Co-authored-by: Pablo Montalvo <39954772+molbap@users.noreply.github.com> Co-authored-by: Keyun Tong <tongkeyun@gmail.com> Co-authored-by: Zijing Liu <liuzijing2014@users.noreply.github.com> Co-authored-by: Lu Fang <fanglu@fb.com> Co-authored-by: Zijing Liu <liuzijing2014@gmail.com> Co-authored-by: Jon Swenson <jmswen@gmail.com> Co-authored-by: jmswen <jmswen@users.noreply.github.com> Co-authored-by: MekkCyber <mekk.cyber@gmail.com> Co-authored-by: Mohamed Mekkouri <93391238+MekkCyber@users.noreply.github.com> Co-authored-by: Mohit Sharma <mohit21sharma.ms@gmail.com> Co-authored-by: Yong Hoon Shin <yhshin@meta.com> Co-authored-by: Marc Sun <marc@huggingface.co> Co-authored-by: drisspg <drisspguessous@gmail.com> Co-authored-by: Cyril Vallez <cyril.vallez@gmail.com> Co-authored-by: Daniël de Kok <me@danieldk.eu> Co-authored-by: Lysandre <hi@lysand.re> Co-authored-by: Ye (Charlotte) Qi <ye.charlotte.qi@gmail.com> Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
This commit is contained in:
Arthur
0
tests/models/llama4/__init__.py
Normal file
0
tests/models/llama4/__init__.py
Normal file
128
tests/models/llama4/test_image_processing_llama4.py
Normal file
128
tests/models/llama4/test_image_processing_llama4.py
Normal file
@@ -0,0 +1,128 @@
|
||||
# coding=utf-8
|
||||
# Copyright 2022 HuggingFace Inc.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
|
||||
|
||||
import unittest
|
||||
|
||||
from transformers.testing_utils import require_torch, require_vision
|
||||
from transformers.utils import is_torch_available, is_torchvision_available, is_vision_available
|
||||
|
||||
from ...test_image_processing_common import ImageProcessingTestMixin, prepare_image_inputs
|
||||
|
||||
|
||||
if is_torch_available():
|
||||
pass
|
||||
|
||||
if is_vision_available() and is_torchvision_available():
|
||||
from transformers import Llama4ImageProcessorFast
|
||||
|
||||
|
||||
class Llama4ImageProcessingTester(unittest.TestCase):
|
||||
def __init__(
|
||||
self,
|
||||
parent,
|
||||
batch_size=7,
|
||||
num_channels=3,
|
||||
image_size=18,
|
||||
min_resolution=30,
|
||||
max_resolution=400,
|
||||
max_patches=1,
|
||||
do_resize=True,
|
||||
size=None,
|
||||
do_normalize=True,
|
||||
do_pad=False,
|
||||
image_mean=[0.5, 0.5, 0.5],
|
||||
image_std=[0.5, 0.5, 0.5],
|
||||
do_convert_rgb=True,
|
||||
):
|
||||
super().__init__()
|
||||
size = size if size is not None else {"height": 20, "width": 20}
|
||||
self.parent = parent
|
||||
self.batch_size = batch_size
|
||||
self.num_channels = num_channels
|
||||
self.image_size = image_size
|
||||
self.min_resolution = min_resolution
|
||||
self.max_resolution = max_resolution
|
||||
self.max_patches = max_patches
|
||||
self.do_resize = do_resize
|
||||
self.size = size
|
||||
self.do_normalize = do_normalize
|
||||
self.image_mean = image_mean
|
||||
self.image_std = image_std
|
||||
self.do_pad = do_pad
|
||||
self.do_convert_rgb = do_convert_rgb
|
||||
|
||||
def prepare_image_processor_dict(self):
|
||||
return {
|
||||
"max_patches": self.max_patches,
|
||||
"do_resize": self.do_resize,
|
||||
"size": self.size,
|
||||
"do_normalize": self.do_normalize,
|
||||
"image_mean": self.image_mean,
|
||||
"image_std": self.image_std,
|
||||
"do_convert_rgb": self.do_convert_rgb,
|
||||
"do_pad": self.do_pad,
|
||||
}
|
||||
|
||||
def expected_output_image_shape(self, images):
|
||||
return self.num_channels, self.size["height"], self.size["width"]
|
||||
|
||||
def prepare_image_inputs(self, equal_resolution=False, numpify=False, torchify=False):
|
||||
return prepare_image_inputs(
|
||||
batch_size=self.batch_size,
|
||||
num_channels=self.num_channels,
|
||||
min_resolution=self.min_resolution,
|
||||
max_resolution=self.max_resolution,
|
||||
equal_resolution=equal_resolution,
|
||||
numpify=numpify,
|
||||
torchify=torchify,
|
||||
)
|
||||
|
||||
|
||||
@require_torch
|
||||
@require_vision
|
||||
class Llama4ImageProcessingTest(ImageProcessingTestMixin, unittest.TestCase):
|
||||
test_slow_image_processor = False
|
||||
fast_image_processing_class = Llama4ImageProcessorFast if is_torchvision_available() else None
|
||||
|
||||
def setUp(self):
|
||||
super().setUp()
|
||||
self.image_processor_tester = Llama4ImageProcessingTester(self)
|
||||
|
||||
@property
|
||||
def image_processor_dict(self):
|
||||
return self.image_processor_tester.prepare_image_processor_dict()
|
||||
|
||||
def test_image_processor_properties(self):
|
||||
for image_processing_class in self.image_processor_list:
|
||||
image_processor = image_processing_class(**self.image_processor_dict)
|
||||
self.assertTrue(hasattr(image_processor, "do_resize"))
|
||||
self.assertTrue(hasattr(image_processor, "size"))
|
||||
self.assertTrue(hasattr(image_processor, "do_normalize"))
|
||||
self.assertTrue(hasattr(image_processor, "image_mean"))
|
||||
self.assertTrue(hasattr(image_processor, "image_std"))
|
||||
self.assertTrue(hasattr(image_processor, "do_convert_rgb"))
|
||||
|
||||
def test_split_tiles(self):
|
||||
for image_processing_class in self.image_processor_list:
|
||||
image_processor = image_processing_class(**self.image_processor_dict)
|
||||
image = self.image_processor_tester.prepare_image_inputs(equal_resolution=True)[0]
|
||||
processed_images = image_processor(
|
||||
image,
|
||||
max_patches=16,
|
||||
)
|
||||
self.assertEqual(len(processed_images.pixel_values), 1)
|
||||
self.assertEqual(processed_images.pixel_values[0].shape[0], 17)
|
||||
self.assertEqual(processed_images.pixel_values[0].shape[-2:], (20, 20))
|
||||
121
tests/models/llama4/test_modeling_llama4.py
Normal file
121
tests/models/llama4/test_modeling_llama4.py
Normal file
@@ -0,0 +1,121 @@
|
||||
# coding=utf-8
|
||||
# Copyright 2025 The HuggingFace Inc. team. All rights reserved.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
"""Testing suite for the PyTorch Llama4 model."""
|
||||
|
||||
import unittest
|
||||
|
||||
from transformers import is_torch_available
|
||||
from transformers.testing_utils import (
|
||||
require_read_token,
|
||||
require_torch_large_gpu,
|
||||
slow,
|
||||
torch_device,
|
||||
)
|
||||
|
||||
|
||||
if is_torch_available():
|
||||
import torch
|
||||
|
||||
from transformers import (
|
||||
Llama4ForConditionalGeneration,
|
||||
Llama4Processor,
|
||||
)
|
||||
|
||||
|
||||
@slow
|
||||
@require_torch_large_gpu
|
||||
@require_read_token
|
||||
class Llama4IntegrationTest(unittest.TestCase):
|
||||
model_id = "ll-re/Llama-4-17B-Omni-Instruct"
|
||||
# This variable is used to determine which CUDA device are we using for our runners (A10 or T4)
|
||||
# Depending on the hardware we get different logits / generations
|
||||
cuda_compute_capability_major_version = None
|
||||
|
||||
@classmethod
|
||||
def setUpClass(cls):
|
||||
if is_torch_available() and torch.cuda.is_available():
|
||||
# 8 is for A100 / A10 and 7 for T4
|
||||
cls.cuda_compute_capability_major_version = torch.cuda.get_device_capability()[0]
|
||||
cls.model = Llama4ForConditionalGeneration.from_pretrained(
|
||||
"ll-re/Llama-4-17B-Omni-Instruct", device_map="auto", torch_dtype=torch.float32
|
||||
)
|
||||
|
||||
def setUp(self):
|
||||
self.processor = Llama4Processor.from_pretrained("ll-re/Llama-4-17B-Omni-Instruct", padding_side="left")
|
||||
|
||||
url = "https://huggingface.co/datasets/hf-internal-testing/fixtures-captioning/resolve/main/cow_beach_1.png"
|
||||
self.messages = [
|
||||
{"role": "system", "content": [{"type": "text", "text": "You are a helpful assistant."}]},
|
||||
{
|
||||
"role": "user",
|
||||
"content": [
|
||||
{"type": "image", "url": url},
|
||||
{"type": "text", "text": "What is shown in this image?"},
|
||||
],
|
||||
},
|
||||
]
|
||||
|
||||
def test_model_17b_16e_fp16(self):
|
||||
EXPECTED_TEXT = [
|
||||
"The capital of France is Paris, which is located in the north-central part of the country. Paris is known for its iconic landmarks such as the",
|
||||
"Roses are red, violets are blue, and this poem is about you. Roses are red, violets are blue, and I love",
|
||||
]
|
||||
|
||||
messages = [
|
||||
{"role": "user", "content": "Who are you?"},
|
||||
]
|
||||
inputs = self.processor.apply_chat_template(
|
||||
messages, add_generation_prompt=True, return_tensors="pt", return_dict=True
|
||||
).to(torch_device)
|
||||
|
||||
output = self.model.generate(**inputs, max_new_tokens=100)
|
||||
output_text = self.processor.batch_decode(output, skip_special_tokens=True)
|
||||
|
||||
print(output_text)
|
||||
self.assertEqual(output_text, EXPECTED_TEXT)
|
||||
|
||||
def test_model_17b_16e_batch(self):
|
||||
messages_2 = [
|
||||
{"role": "system", "content": [{"type": "text", "text": "You are a helpful assistant."}]},
|
||||
{
|
||||
"role": "user",
|
||||
"content": [
|
||||
{
|
||||
"type": "image",
|
||||
"url": "https://huggingface.co/datasets/hf-internal-testing/fixtures-captioning/resolve/main/cow_beach_1.png",
|
||||
},
|
||||
{"type": "image", "url": "https://www.ilankelman.org/stopsigns/australia.jpg"},
|
||||
{"type": "text", "text": "Are these images identical?"},
|
||||
],
|
||||
},
|
||||
]
|
||||
|
||||
inputs = self.processor.apply_chat_template(
|
||||
[self.messages, messages_2],
|
||||
tokenize=True,
|
||||
return_dict=True,
|
||||
return_tensors="pt",
|
||||
padding=True,
|
||||
add_generation_prompt=True,
|
||||
).to(torch_device)
|
||||
|
||||
output = self.model.generate(**inputs, max_new_tokens=30, do_sample=False)
|
||||
output_text = self.processor.batch_decode(output, skip_special_tokens=True)
|
||||
|
||||
EXPECTED_TEXTS = [
|
||||
'user\nYou are a helpful assistant.\n\n\n\n\n\nWhat is shown in this image?\nmodel\nCertainly! \n\nThe image shows a brown cow standing on a sandy beach with clear turquoise water and a blue sky in the background. It looks like',
|
||||
"user\nYou are a helpful assistant.\n\n\n\n\n\n\n\n\n\nAre these images identical?\nmodel\nNo, these images are not identical. \n\nHere's a breakdown of the differences:\n\n* **Image 1:** Shows a cow"
|
||||
] # fmt: skip
|
||||
self.assertEqual(output_text, EXPECTED_TEXTS)
|
||||
65
tests/models/llama4/test_processor_llama4.py
Normal file
65
tests/models/llama4/test_processor_llama4.py
Normal file
@@ -0,0 +1,65 @@
|
||||
# Copyright 2024 The HuggingFace Team. All rights reserved.
|
||||
#
|
||||
# Licensed under the Apache License, Version 2.0 (the "License");
|
||||
# you may not use this file except in compliance with the License.
|
||||
# You may obtain a copy of the License at
|
||||
#
|
||||
# http://www.apache.org/licenses/LICENSE-2.0
|
||||
#
|
||||
# Unless required by applicable law or agreed to in writing, software
|
||||
# distributed under the License is distributed on an "AS IS" BASIS,
|
||||
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||||
# See the License for the specific language governing permissions and
|
||||
# limitations under the License.
|
||||
|
||||
import shutil
|
||||
import tempfile
|
||||
import unittest
|
||||
from typing import Optional
|
||||
|
||||
from transformers import AutoProcessor, Llama4Processor, PreTrainedTokenizerFast
|
||||
from transformers.testing_utils import require_vision
|
||||
from transformers.utils import is_vision_available
|
||||
|
||||
from ...test_processing_common import ProcessorTesterMixin
|
||||
|
||||
|
||||
if is_vision_available():
|
||||
from transformers import Llama4ImageProcessorFast
|
||||
|
||||
|
||||
@require_vision
|
||||
class Llama4ProcessorTest(ProcessorTesterMixin, unittest.TestCase):
|
||||
processor_class = Llama4Processor
|
||||
|
||||
def setUp(self):
|
||||
self.tmpdirname = tempfile.mkdtemp()
|
||||
|
||||
image_processor = Llama4ImageProcessorFast(max_patches=1, size={"height": 20, "width": 20})
|
||||
tokenizer = PreTrainedTokenizerFast.from_pretrained("unsloth/Llama-3.2-11B-Vision-Instruct-unsloth-bnb-4bit")
|
||||
processor_kwargs = self.prepare_processor_dict()
|
||||
processor = Llama4Processor(image_processor, tokenizer, **processor_kwargs)
|
||||
processor.save_pretrained(self.tmpdirname)
|
||||
|
||||
def get_tokenizer(self, **kwargs):
|
||||
return AutoProcessor.from_pretrained(self.tmpdirname, **kwargs).tokenizer
|
||||
|
||||
def get_image_processor(self, **kwargs):
|
||||
return AutoProcessor.from_pretrained(self.tmpdirname, **kwargs).image_processor
|
||||
|
||||
def tearDown(self):
|
||||
shutil.rmtree(self.tmpdirname)
|
||||
|
||||
# Override as Llama4ProcessorProcessor needs image tokens in prompts
|
||||
def prepare_text_inputs(self, batch_size: Optional[int] = None):
|
||||
if batch_size is None:
|
||||
return "lower newer <image>"
|
||||
|
||||
if batch_size < 1:
|
||||
raise ValueError("batch_size must be greater than 0")
|
||||
|
||||
if batch_size == 1:
|
||||
return ["lower newer <image>"]
|
||||
return ["lower newer <image>", "<image> upper older longer string"] + ["<image> lower newer"] * (
|
||||
batch_size - 2
|
||||
)
|
||||
Reference in New Issue
Block a user