whatmodel.sh / MiniChat-1.5-3B

MiniChat-1.5-3B

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

MiniChat-1.5-3B is an open-weights language model from GeneZC, released November 2023. Its published size is 3.02B parameters. Licence: apache-2.0.

Catalogue identity notes

Maker repo GeneZC/MiniChat-1.5-3B @91b66d3: licence tag apache-2.0. 3,020,442,624 parameters from the Hugging Face safetensors metadata (dense, so active = total). Released 2023-11: month the maker's Hugging Face repo was created (createdAt 2023-11-26).

33.2±2.3#826 of 949 LLMsfrom 2 benchmark results across 1 source* thinly tested, treat the rank as provisional

$ how it compares

The LLMs ranked either side of MiniChat-1.5-3B, out of 1132 with a score, then more from GeneZC.

#modelscorewhereRAMreleased
994StripedHyena-Nous-7B Together AI33±2comparison only2023-12
995Granite 3B Code Base (2K) IBM33±1local~3.7GB2024-04
996MiniChat-1.5-3B GeneZC33±2local~3.4GB2023-11
997GPT-3 davinci (API base model) OpenAI33±1comparison only2020-06
998StarCoderBase 15.5B BigCode33±1local~12GB2023-05
more from GeneZC
1047MiniChat-3B GeneZC29±3local~3.4GB2023-11

see all LLMs · find the ones that fit your machine

$ what it's good at

chat & writing▁▁▁▄▆▅▆█▆▆▆▂ #433 of 466 * · 98% of its score · 2 tests

Beats Code Llama 70B Instruct, Phi-2 (2.7B), Qwen2.5 1.5B Instruct here, though they rank higher overall. Trails Gemma 3 270M IT, SmolLM2 360M Instruct, Granite 4.0 350M here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
AlpacaEval 2 length-controlled7.7% ✎▁▂▂▃▆▆█▆▄▄▁▁ beats 17/122medium · avg 39%▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 72%typical
AlpacaEval 1.0 win rate (vs text-davinci-003, GPT-4 judge)78.5% ✎▂▂·▆·▇▃▆█▆▄▂ beats 18/45very easy · avg 96%▇▇▇▇▇▇▇▇▇▇ 25%sharp · at its level · noisy
other tests1 not in a task · 0.0% of its score
testresultvs the others in its rundifficultycountswhy
Open PT-LLM Leaderboard average (9 Portuguese tasks)0.0% ✎▃▁▁▁▁▁▁▁▂█▅▁ beats 0/276medium · avg 73%—failed run: All nine published task scores are 0.0 in this Open PT row (failed run with no parsed answers), not a measurement of ability.

$ where to run it

llama.cpp · recommended

version

  1. Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
    macOS / Linux
    curl -LsSf https://llama.app/install.sh | sh
    Windows
    irm https://llama.app/install.ps1 | iex

    other ways to install ↗

  2. Run MiniChat-1.5-3B · needs ~1.8 GB · from huggingface.co/mradermacher
    chat in the terminal
    llama cli -hf mradermacher/MiniChat-1.5-3B-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf mradermacher/MiniChat-1.5-3B-GGUF:Q4_K_M

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run MiniChat-1.5-3B · needs ~1.8 GB · from huggingface.co/mradermacher

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/mradermacher/MiniChat-1.5-3B-GGUF@Q4_K_M -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run MiniChat-1.5-3B · needs ~1.8 GB · from huggingface.co/mradermacher
    chat in the terminal
    ollama run hf.co/mradermacher/MiniChat-1.5-3B-GGUF:Q4_K_M

Other ways to run it

  1. Run it as published · needs ~6.04 GB · from huggingface.co/GeneZC (official)
    uv run --with torch --with 'transformers<5' --with accelerate python -c "from transformers import pipeline; print(pipeline('text-generation', model='GeneZC/MiniChat-1.5-3B', torch_dtype='auto', device_map='auto')('Once upon a time', max_new_tokens=40)[0]['generated_text'])"

Official weights and documentation: huggingface.co/GeneZC ↗

$ how the score was worked out · AlpacaEval 1.0 and 2.0 leader… 98%

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
AlpacaEval 1.0 and 2.0 leaderboards (tatsu-lab/alpaca_eval repository)2▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 98%Apache-2.0
prior · pull toward the middle▇ 2%
weighted average= 33.2

not counted (1)

Open Portuguese LLM Leaderboard (archived)Open PT-LLM Leaderboard average (9 Portuguese tasks) [correct]0.0%failed run: All nine published task scores are 0.0 in this Open PT row (failed run with no parsed answers), not a measurement of ability.

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json