bartowski/h2o-danube2-1.8b-chat-GGUF

Llamacpp Quantizations of h2o-danube2-1.8b-chat

Using llama.cpp release b2589 for quantization.

Original model: https://huggingface.co/h2oai/h2o-danube2-1.8b-chat

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

Filename	Quant type	File Size	Description
h2o-danube2-1.8b-chat-Q8_0.gguf	Q8_0	1.94GB	Extremely high quality, generally unneeded but max available quant.
h2o-danube2-1.8b-chat-Q6_K.gguf	Q6_K	1.50GB	Very high quality, near perfect, recommended.
h2o-danube2-1.8b-chat-Q5_K_M.gguf	Q5_K_M	1.30GB	High quality, very usable.
h2o-danube2-1.8b-chat-Q5_K_S.gguf	Q5_K_S	1.27GB	High quality, very usable.
h2o-danube2-1.8b-chat-Q5_0.gguf	Q5_0	1.27GB	High quality, older format, generally not recommended.
h2o-danube2-1.8b-chat-Q4_K_M.gguf	Q4_K_M	1.11GB	Good quality, uses about 4.83 bits per weight.
h2o-danube2-1.8b-chat-Q4_K_S.gguf	Q4_K_S	1.05GB	Slightly lower quality with small space savings.
h2o-danube2-1.8b-chat-IQ4_NL.gguf	IQ4_NL	1.06GB	Decent quality, similar to Q4_K_S, new method of quanting,
h2o-danube2-1.8b-chat-IQ4_XS.gguf	IQ4_XS	1.01GB	Decent quality, new method with similar performance to Q4.
h2o-danube2-1.8b-chat-Q4_0.gguf	Q4_0	1.05GB	Decent quality, older format, generally not recommended.
h2o-danube2-1.8b-chat-Q3_K_L.gguf	Q3_K_L	.98GB	Lower quality but usable, good for low RAM availability.
h2o-danube2-1.8b-chat-Q3_K_M.gguf	Q3_K_M	.90GB	Even lower quality.
h2o-danube2-1.8b-chat-IQ3_M.gguf	IQ3_M	.85GB	Medium-low quality, new method with decent performance.
h2o-danube2-1.8b-chat-IQ3_S.gguf	IQ3_S	.82GB	Lower quality, new method with decent performance, recommended over Q3 quants.
h2o-danube2-1.8b-chat-Q3_K_S.gguf	Q3_K_S	.82GB	Low quality, not recommended.
h2o-danube2-1.8b-chat-Q2_K.gguf	Q2_K	.71GB	Extremely low quality, not recommended.

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