T-lite-it-1.0-Q8_0-GGUF

🚨 T-lite is designed for further fine-tuning and is not intended as a ready-to-use conversational assistant. Users are advised to exercise caution and are responsible for any additional training and oversight required to ensure the model's responses meet acceptable ethical and safety standards. The responsibility for incorporating this model into industrial or commercial solutions lies entirely with those who choose to deploy it.

Description

This repository contains the T-lite-it-1.0 model, which has been quantized into the GGUF format using the llama.cpp repository.

📊 Benchmarks

Detailed evaluation results of oringal model can be found in our habr post.

Benchmark	T-lite-it-1.0	T-lite-it-1.0-Q8_0
Arena-Hard-Ru	64.38 (-2.1, 2.5)	64.21 (-2.2, 2.7)

Llama.cpp usage

Server

From HF:

llama-server --hf-repo t-tech/T-lite-it-1.0-Q8_0-GGUF --hf-file t-lite-it-1.0-q8_0.gguf -c 8192

Or locally:

./build/bin/llama-server -m t-lite-it-1.0-q8_0.gguf -c 8192

POST

curl --request POST \
    --url http://localhost:8080/completion \
    --header "Content-Type: application/json" \
    --data '{
        "prompt": "<|im_start|>user\nРасскажи мне чем отличается Python от C++?\n<|im_end|>\n<|im_start|>assistant\n",
        "n_predict": 256
    }'

ollama usage

Serve

ollama serve

Run

From HF:

ollama run hf.co/t-tech/T-lite-it-1.0-Q8_0-GGUF:Q8_0 "Расскажи мне про отличия C++ и Python"

Or locally:

ollama create example -f Modelfile
ollama run example "Расскажи мне про отличия C++ и Python"

where Modelfile is

FROM ./t-lite-it-1.0-q8_0.gguf