Qwen3-1.7B-GPTQ-W4A16_gsm8k2048 / quantization_config.yaml
chieunq's picture
Upload folder using huggingface_hub
dafa99b verified
Raw
History Blame Contribute Delete
358 Bytes
dataset:
calibration:
max_seq_length: 2048
name: openai/gsm8k
num_samples: 2048
seed: 42
subset: main
model:
model_id: Qwen/Qwen3-1.7B
torch_dtype: auto
output:
log_dir: sparse_logs
output_path: result
save_compressed: true
save_dir: null
quantization:
ignore:
- lm_head
method: gptq
scheme: W4A16
targets: Linear