Modelpublic

mradermacher/falcon-40b-instruct-i1-GGUF

sourceHugging Faceapache-2.0updated 2y agoView on Hugging Face

1likes7kdownloads

Model Card

weighted/imatrix quants of https://huggingface.co/tiiuae/falcon-40b-instruct

static quants are available at https://huggingface.co/mradermacher/falcon-40b-instruct-GGUF

If you are unsure how to use GGUF files, refer to one of TheBloke's READMEs for more details, including on how to concatenate multi-part files.

(sorted by size, not necessarily quality. IQ-quants are often preferable over similar sized non-IQ quants)

Link	Type	Size/GB	Notes
GGUF	i1-IQ1_S	9.8	for the desperate
GGUF	i1-IQ1_M	10.1	mostly desperate
GGUF	i1-IQ2_XXS	12.0
GGUF	i1-IQ2_XS	13.2
GGUF	i1-IQ2_S	14.0
GGUF	i1-IQ2_M	15.1
GGUF	i1-Q2_K	16.4	IQ3_XXS probably better
GGUF	i1-IQ3_XXS	17.0	lower quality
GGUF	i1-IQ3_XS	18.4
GGUF	i1-IQ3_S	18.9	beats Q3_K*
GGUF	i1-Q3KS	18.9	IQ3_XS probably better
GGUF	i1-IQ3_M	19.8
GGUF	i1-Q3KM	20.7	IQ3_S probably better
GGUF	i1-Q3KL	22.2	IQ3_M probably better
GGUF	i1-IQ4_XS	22.6
GGUF	i1-Q4_0	24.0	fast, low quality
GGUF	i1-Q4KS	24.4	optimal size/speed/quality
GGUF	i1-Q4KM	26.1	fast, recommended
GGUF	i1-Q5KS	29.6
GGUF	i1-Q5KM	31.2
GGUF	i1-Q6_K	35.1	practically like static Q6_K

Here is a handy graph by ikawrakow comparing some lower-quality quant types (lower is better):

And here are Artefact2's thoughts on the matter: https://gist.github.com/Artefact2/b5f810600771265fc1e39442288e8ec9

See https://huggingface.co/mradermacher/model_requests for some answers to questions you might have and/or if you want some other model quantized.

I thank my company, nethype GmbH, for letting me use its servers and providing upgrades to my workstation to enable this work in my free time.