ModelÜberprüftExtern

Qwen3 4B Thinking 2507 Llmc Awq Calib Openr1 Math 220k All Train W4a16 G128 N256 S1024

method: awq weight format: W4A16 group size: 128 calibration dataset: open-r1/OpenR1-Math-220k calibration dataset config: all calibration split: train calibration samples: 256 max sequence length: 1024

Quellehuggingface.co/meteorain/Qwen3-4B-Thinking-2507-llmc-awq-calib-openr1-math-220k-all-train-w4a16-g128-n256-s1024/tree/78304ea2207b7ae104a3d03434a2031d3cfacab5

ArtText modelBasismodellQwen/Qwen3-4B-Thinking-2507VersionmodelVerleger@meteorainCBewertung

Modellquelle

Art: Text model
Dateigröße: 2,48 GB
Basismodell: Qwen/Qwen3-4B-Thinking-2507
Version: vmodel
Aktualisiert: 21. Okt. 2026
Verleger: @meteorain
Downloads: 13
Parameter: 4B
Quelle: Hugging Face

Quellenbeschreibung

Qwen3-4B-Thinking-2507-llmc-awq-calib-openr1-math-220k-all-train-w4a16-g128-n256-s1024

Base model: Qwen/Qwen3-4B-Thinking-2507

Quantized with llm-compressor.

method: awq
weight format: W4A16
group size: 128
calibration dataset: open-r1/OpenR1-Math-220k
calibration dataset config: all
calibration split: train
calibration samples: 256
max sequence length: 1024

Quellen

1 Quelle

Hugging FacePrimärVersion 78304ea2207b7ae104a3d03434a2031d3cfacab5

huggingface.co/meteorain/Qwen3-4B-Thinking-2507-llmc-awq-calib-openr1-math-220k-all-train-w4a16-g128-n256-s1024/tree/78304ea2207b7ae104a3d03434a2031d3cfacab5

Verifiziert 21. Okt.

Modellartefakte

1 Artefakt

model.safetensors

safetensors · 2,48 GB · SHA-256 52ec9f654d7a…be32 · Hugging Face

Herunterladen

Quellenauszüge

3 Auszüge

Readme11. Julihuggingface.co/meteorain/Qwen3-4B-Thinking-2507-llmc-awq-calib-openr1-math-220k-all-train-w4a16-g128-n256-s1024

--- library_name: transformers base_model: - Qwen/Qwen3-4B-Thinking-2507 base_model_relation: quantized pipeline_tag: text-generation tags: - Qwen3 - AWQ - W4A16 - llm-compressor --- # Qwen3-4B-Thinking-2507-llmc-awq-calib-openr1-math-220k-all-train-w4a16-g128-n256-s1024 Base model: `Qwen/Qwen3-4B-Thinking-2507` Quantized with `llm-compressor`. - method: `awq` - weight format: `W4A16` - group size: `128` - calibration dataset: `open-r1/OpenR1-Math-220k` - calibration dataset config: `all` - calibration split: `train` - calibration samples: `256` - max sequence length: `1024`

Note21. Okt.huggingface.co/meteorain/Qwen3-4B-Thinking-2507-llmc-awq-calib-openr1-math-220k-all-train-w4a16-g128-n256-s1024

Source context: 13 downloads · 0 likes · Pipeline text-generation · Library transformers · Repo meteorain/Qwen3-4B-Thinking-2507-llmc-awq-calib-openr1-math-220k-all-train-w4a16-g128-n256-s1024

Note18. Julihuggingface.co/meteorain/Qwen3-4B-Thinking-2507-llmc-awq-calib-openr1-math-220k-all-train-w4a16-g128-n256-s1024

Source context: 42 downloads · 0 likes · Pipeline text-generation · Library transformers · Repo meteorain/Qwen3-4B-Thinking-2507-llmc-awq-calib-openr1-math-220k-all-train-w4a16-g128-n256-s1024