Qwen3-1.7B-AWQ-W4A16_ASYM_gsm8k1024 / quantization_config.yaml
chieunq's picture
Upload folder using huggingface_hub
2e93b0a verified
dataset:
calibration:
max_seq_length: 2048
name: openai/gsm8k
num_samples: 1024
seed: 42
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: awq
scheme: W4A16_ASYM
targets: Linear