Warning: something seems to have gone wrong with conversion, output is non-sensical, investigating
Llamacpp Quantizations of Cerebrum-1.0-8x7b
Using llama.cpp release b2440 for quantization.
Original model: https://huggingface.co/AetherResearch/Cerebrum-1.0-8x7b
Download a file (not the whole branch) from below:
Filename | Quant type | File Size | Description |
---|---|---|---|
Cerebrum-1.0-8x7b-Q8_0.gguf | Q8_0 | 49.62GB | Extremely high quality, generally unneeded but max available quant. |
Cerebrum-1.0-8x7b-Q6_K.gguf | Q6_K | 38.37GB | Very high quality, near perfect, recommended. |
Cerebrum-1.0-8x7b-Q5_K_M.gguf | Q5_K_M | 33.22GB | High quality, very usable. |
Cerebrum-1.0-8x7b-Q5_K_S.gguf | Q5_K_S | 32.22GB | High quality, very usable. |
Cerebrum-1.0-8x7b-Q5_0.gguf | Q5_0 | 32.22GB | High quality, older format, generally not recommended. |
Cerebrum-1.0-8x7b-Q4_K_M.gguf | Q4_K_M | 28.44GB | Good quality, similar to 4.25 bpw. |
Cerebrum-1.0-8x7b-Q4_K_S.gguf | Q4_K_S | 26.74GB | Slightly lower quality with small space savings. |
Cerebrum-1.0-8x7b-IQ4_NL.gguf | IQ4_NL | 26.74GB | Good quality, similar to Q4_K_S, new method of quanting, |
Cerebrum-1.0-8x7b-IQ4_XS.gguf | IQ4_XS | 25.37GB | Decent quality, new method with similar performance to Q4. |
Cerebrum-1.0-8x7b-Q4_0.gguf | Q4_0 | 26.44GB | Decent quality, older format, generally not recommended. |
Cerebrum-1.0-8x7b-IQ3_M.gguf | IQ3_M | 21.42GB | Medium-low quality, new method with decent performance. |
Cerebrum-1.0-8x7b-IQ3_S.gguf | IQ3_S | 20.43GB | Lower quality, new method with decent performance, recommended over Q3 quants. |
Cerebrum-1.0-8x7b-Q3_K_L.gguf | Q3_K_L | 24.16GB | Lower quality but usable, good for low RAM availability. |
Cerebrum-1.0-8x7b-Q3_K_M.gguf | Q3_K_M | 22.54GB | Even lower quality. |
Cerebrum-1.0-8x7b-Q3_K_S.gguf | Q3_K_S | 20.43GB | Low quality, not recommended. |
Cerebrum-1.0-8x7b-Q2_K.gguf | Q2_K | 17.30GB | Extremely low quality, not recommended. |
Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski
- Downloads last month
- 192
Model tree for bartowski/Cerebrum-1.0-8x7b-GGUF-old
Base model
mistralai/Mixtral-8x7B-v0.1
Finetuned
mistralai/Mixtral-8x7B-Instruct-v0.1
Quantized
this model