tikeape/Qwen3.5-4B-Claude-Opus-Reasoning-Distill-GGUF

Qwen3.5-4B-Claude-Opus-Reasoning-Distill GGUF

GGUF conversions and quantized builds of Qwen3.5-4B-Claude-Opus-Reasoning-Distill, generated from TeichAI/Qwen3.5-4B-Claude-Opus-Reasoning-Distill using llama.cpp.

Included files

File	Quantization
Qwen3.5-4B-Claude-Opus-Reasoning-Distill-gguf-f16.gguf	f16
Qwen3.5-4B-Claude-Opus-Reasoning-Distill-gguf-hyb-hybrid-1-b-iq1-s-o-q4-k-e-q4-k-t2.gguf	HYBRID (hybrid-1)

Hybrid recipes

hybrid-1

• File: Qwen3.5-4B-Claude-Opus-Reasoning-Distill-gguf-hyb-hybrid-1-b-iq1-s-o-q4-k-e-q4-k-t2.gguf
• Base quant: IQ1_S
• Output tensor override: Q4_K
• Token embedding override: Q4_K
• Tensor overrides:
  • attn_q -> Q4_K
  • attn_v -> Q4_K

Notes

• Filenames follow the pattern base-gguf-<quant>.gguf for standard quants.
• Hybrid filenames follow the pattern base-gguf-hyb-<recipe>-b-<base>-o-<out>-e-<emb>-t<count>.gguf.
• Standard quantizations included: f16 only.
• Hybrid builds included: hybrid-1.