Downloads · 30 days
10
28% of all-time downloads
paligemma-3b-mix-224-FP8-dynamic
nm-testing/paligemma-3b-mix-224-FP8-dynamic
paligemma-3b-mix-224-FP8-dynamic is a machine learning model from nm-testing. Use it for the machine learning task on the model card, and read the license before you ship it in a product.
All-time downloads
36
Public
Parameters
3.5B
7.9 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors7.9 GB · 100%
Parameter types
How the weights are stored.
F8_E4M32B · 57%
Base models
Model card
From the Hugging Face model README
Creation
from transformers import AutoProcessor, PaliGemmaForConditionalGeneration
from llmcompressor.modifiers.quantization import QuantizationModifier
from llmcompressor.transformers import oneshot, wrap_hf_model_class
MODEL_ID = "google/paligemma-3b-mix-224"
# Load model.
model_class = wrap_hf_model_class(PaliGemmaForConditionalGeneration)
model = model_class.from_pretrained(MODEL_ID, device_map="auto", torch_dtype="auto", trust_remote_code=True)
processor = AutoProcessor.from_pretrained(MODEL_ID, trust_remote_code=True)
# Configure the quantization algorithm and scheme.
# In this case, we:
# * quantize the weights to fp8 with per channel via ptq
# * quantize the activations to fp8 with dynamic per token
recipe = QuantizationModifier(
targets="Linear",
scheme="FP8_DYNAMIC",
ignore=["re:.*lm_head", "re:multi_modal_projector.*", "re:vision_tower.*"],
)
# Apply quantization and save to disk in compressed-tensors format.
SAVE_DIR = MODEL_ID.split("/")[1] + "-FP8-dynamic"
oneshot(model=model, recipe=recipe, output_dir=SAVE_DIR)
processor.save_pretrained(SAVE_DIR)
- Type
- paligemma
- Created
- Oct 29, 2024
- Updated
- Oct 29, 2024