mistrallite:7b-v0.1-fp16

61.3K 2 years ago

MistralLite is a fine-tuned model based on Mistral with enhanced capabilities of processing long contexts.

7b

2 years ago

a6c59d49e2a9 · 14GB ·

model
llama
·
7.24B
·
F16
params
{ "num_ctx": 32768, "stop": [ "<|prompter|>", "<|assistant|>", "</s>
template
<|prompter|>{{ .System }} {{ .Prompt }}</s><|assistant|>

Readme

Mistrallite is a fine-tuned model based on Mistral, with enhanced capabilities of processing long context (up to 32K tokens). This model is able to perform significantly better on several long context retrieve and answering tasks.

Reference

Hugging Face