Cygnis Alpha 2 i1
Enhanced Importance Matrix (imatrix) quantizations for Cygnis Alpha 2.
Higher intelligence floor for low-bit inference, preserving complex CoT patterns.
ollama run Simonc-44/Cygnis-Alpha-2-8B-v0.3-i1-GGUF
Why Imatrix?
Standard quantization treats all weights equally. Imatrix (Importance Matrix) uses a calibration dataset to identify which neurons are vital for the model's logic. By protecting these neurons, we achieve 3-bit or 4-bit quants that often match the performance of standard 6-bit files.
Perplexity vs Bitrate (Efficiency)
Visual representation of why Imatrix (i1) quants outperform standard static quants at lower bitrates.
Ollama Config
FROM ./Cygnis-Alpha-2-7B-v0.3.i1-Q4_K_M.ggufTEMPLATE """<|im_start|>system {{ .System }}<|im_end|> <|im_start|>user {{ .Prompt }}<|im_end|> <|im_start|>assistant <|im_thought|> """
PARAMETER stop "<|im_end|>"
Quantized with ❤️ by Simonc-44 | Powered by Llama.cpp Imatrix
- Downloads last month
- 1,778
Model tree for Simonc-44/Cygnis-Alpha-2-8B-v0.3-i1-GGUF
Base model
Simonc-44/Cygnis-Alpha-2-8B-v0.1