mannix/
llava-phi3:iq4_xs

1,926 1 year ago

A new small LLaVA model fine-tuned from Phi 3 Mini [I-Quants]

vision

1 year ago

7c7b97d18d11 · 2.7GB ·

model
llama
·
3.82B
·
IQ1_M
projector
clip
·
303M
·
F16
params
{ "num_ctx": 4096, "num_keep": 4, "stop": [ "<|user|>", "<|assistant|>",
template
{{ if .System }}<|system|> {{ .System }}<|end|> {{ end }}{{ if .Prompt }}<|user|> {{ .Prompt }}<|end

Readme

image.png

llava-phi3 is a LLaVA model fine-tuned from Phi 3 Mini 4k, with strong performance benchmarks on par with the original LLaVA model:

image.png

References

Hugging Face

GitHub