Add GGUF exports for analysis (F16, Q8_0, Q5_K_M, Q4_K_M)
#4
by pylotlight - opened
Adds backbone-only GGUF exports for analysis (discussion #2).
- Files:
gguf/with F16 (7.85 GB), Q8_0 (4.17 GB), Q5_K_M (2.87 GB), Q4_K_M (2.54 GB) +gguf/README.md. - Scope: backbone tensors only (qwen35, 378 tensors incl. SSM). Not a drop-in
Decider.decide()runtime in llama.cpp (excludesreadout.safetensors,thinker/, marker-readout protocol). - Conversion: llama.cpp @
3ba1ce0/ gguf 0.19.0 (torch 2.15.0, transformers 5.22.2).
pylotlight changed pull request status to open
pylotlight changed pull request status to closed
pylotlight changed pull request status to open
pylotlight changed pull request title from Add GGUF exports for analysis (Q8_0, Q5_K_M, Q4_K_M) to Add GGUF exports for analysis (F16, Q8_0, Q5_K_M, Q4_K_M)