inference-optimization
/

Qwen3.5-35B-A3B-FP8-Dynamic

qwen3_5_moe_text

compressed-tensors

Model card Files Files and versions

Qwen3.5-35B-A3B-FP8-Dynamic / recipe.yaml

krishnateja95's picture

Upload folder using huggingface_hub

8023837 verified about 2 months ago

history blame contribute delete

248 Bytes

	default_stage:
	default_modifiers:
	QuantizationModifier:
	targets: [Linear]
	ignore: [lm_head, 're:.mlp.gate$', 're:.mlp.shared_expert_gate$', 're:.linear_attn.']
	scheme: FP8_DYNAMIC
	bypass_divisibility_checks: false