Views
No views yet
| File | Quant | Size |
|---|---|---|
evomodel-1.9c-ck845-Q8_0.gguf | Q8_0 | ~8.9 GB |
evomodel-1.9c-ck845-NVFP4.gguf | NVFP4 | ~5.7 GB |
evomodel-1.9c-ck845-Q4_K_M.gguf | Q4_K_M | ~5.2 GB |
token_embd and output are kept at Q8_0.| Architecture | Qwen-3.5 (9B) |
| Adapter | LoRA r=8, alpha=8, merged at scale 1.0 |
| Context | up to 262k |
1llama-server -m evomodel-1.9c-ck845-Q4_K_M.gguf \
2 -ngl 99 --jinja --reasoning on -c 32768 --host 0.0.0.0 --port 8080<function=…> dialect inside <tool_call>; the serving
client parses the raw text (llama.cpp does not parse this dialect itself).