README.md

metadata

license: apache-2.0
datasets:
  - Azure99/blossom-chat-v3
  - Azure99/blossom-math-v4
  - Azure99/blossom-wizard-v3
  - Azure99/blossom-orca-v3
language:
  - zh
  - en
quantized_by: bartowski
pipeline_tag: text-generation

Llamacpp Quantizations of blossom-v5-9b

Using llama.cpp release b2440 for quantization.

Original model: https://huggingface.co/Azure99/blossom-v5-9b

Download a file (not the whole branch) from below:

Filename	Quant type	File Size	Description
blossom-v5-9b-Q8_0.gguf	Q8_0	9.38GB	Extremely high quality, generally unneeded but max available quant.
blossom-v5-9b-Q6_K.gguf	Q6_K	7.24GB	Very high quality, near perfect, recommended.
blossom-v5-9b-Q5_K_M.gguf	Q5_K_M	6.25GB	High quality, very usable.
blossom-v5-9b-Q5_K_S.gguf	Q5_K_S	6.10GB	High quality, very usable.
blossom-v5-9b-Q5_0.gguf	Q5_0	6.10GB	High quality, older format, generally not recommended.
blossom-v5-9b-Q4_K_M.gguf	Q4_K_M	5.32GB	Good quality, similar to 4.25 bpw.
blossom-v5-9b-Q4_K_S.gguf	Q4_K_S	5.07GB	Slightly lower quality with small space savings.
blossom-v5-9b-IQ4_NL.gguf	IQ4_NL	5.08GB	Good quality, similar to Q4_K_S, new method of quanting,
blossom-v5-9b-IQ4_XS.gguf	IQ4_XS	4.82GB	Decent quality, new method with similar performance to Q4.
blossom-v5-9b-Q4_0.gguf	Q4_0	5.03GB	Decent quality, older format, generally not recommended.
blossom-v5-9b-IQ3_M.gguf	IQ3_M	4.05GB	Medium-low quality, new method with decent performance.
blossom-v5-9b-IQ3_S.gguf	IQ3_S	3.91GB	Lower quality, new method with decent performance, recommended over Q3 quants.
blossom-v5-9b-Q3_K_L.gguf	Q3_K_L	4.69GB	Lower quality but usable, good for low RAM availability.
blossom-v5-9b-Q3_K_M.gguf	Q3_K_M	4.32GB	Even lower quality.
blossom-v5-9b-Q3_K_S.gguf	Q3_K_S	3.89GB	Low quality, not recommended.
blossom-v5-9b-Q2_K.gguf	Q2_K	3.35GB	Extremely low quality, not recommended.

Want to support my work? Visit my ko-fi page here: https://ko-fi.com/bartowski