whatmodel.sh / StarCoderBase 15.5B

StarCoderBase 15.5B

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

StarCoderBase 15.5B is an open-weights language model from BigCode, released May 2023. Its published size is 15.5B parameters. Licence: bigcode-openrail-m.

Catalogue identity notes

Comparison-only. Maker repo bigcode/starcoderbase @88ec578: licence tag bigcode-openrail-m. Parameters: nominal size from the model name (no safetensors metadata in the repo) (dense, so active = total). Released 2023-05: month the maker's Hugging Face repo was created (createdAt 2023-05-03). StarCoderBase 15.5B (May 2023), trained on 80+ languages; the Python-fine-tuned StarCoder is a separate id (starcoder).

33.0±1.0#618 of 719 LLMsfrom 12 benchmark results across 3 sources

$ how it compares

The LLMs ranked either side of StarCoderBase 15.5B, out of 828 with a score, then more from BigCode.

#modelscorewhereRAMreleased
713Granite 3B Code Base (2K) IBM33±1local~3.7GB2024-04
714GPT-3 davinci (API base model) OpenAI33±1comparison only2020-06
715StarCoderBase 15.5B BigCode33±1local~12GB2023-05
716Vicuna 7B v1.5 LMSYS33±1local~5.8GB2023-08
717Guanaco 33B University of Washington33±2local~70GB2023-05
more from BigCode
521StarCoder2-15B-Instruct-v0.1 BigCode46±1local~11GB2024-04
639StarCoder2 15B BigCode39±1local~11GB2024-02
698StarCoder2 7B BigCode35±1local~6.1GB2024-02

see all LLMs · find the ones that fit your machine

$ what it's good at

coding▁▁▁▃▆██▇▆▄▃▁ #419 of 458 · 94% of its score · 8 tests

Beats DeepSeek R1 Distill Qwen 7B, Qwen3.5 2B, DeepSeek R1 Distill Llama 8B here, though they rank higher overall. Trails Llama 3.2 1B Instruct, SmolLM2 1.7B Instruct here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
HumanEval lab29.3% ✎▂··█▂▄▂··▄▂▂ beats 3/12easy · avg 79%▇▇▇▇▇▇▇▇▇▇ 24%sharp · at its level
MBPP +137.4% ✎█·▅▃▃▃▅▃▃▅·▅ beats 6/15easy · avg 82%▇▇▇▇▇▇▇▇ 19%at its level · several runs
HumanEvalSynthesize (HumanEvalPack, 6-language average)31.9% ✎▁·▂▁▂▆▅█▄▄▁▂ beats 10/35medium · avg 64%▇▇▇▇▇ 13%at its level
CRUXEval-I +131.3% ✎▁··▁▅▂█▇▁··▁ beats 4/22medium · avg 59%▇▇▇▇ 9.2%at its level · several runs
CRUXEval-O +134.2% ✎▁·▂▃▁█▁▂···▁ beats 5/22medium · avg 55%▇▇▇ 8.1%blunt · at its level · several runs
DS-1000 lab23.8% ✎▄▄·▄██▄█·█·▄ beats 5/12medium · avg 52%▇▇▇ 7.9%typical
HumanEvalExplain (HumanEvalPack, 6-language average)15.3% ✎▄▇█▃▄▇▄▁▇▃·▁ beats 11/35medium · avg 49%▇▇▇ 6.8%typical
HumanEvalFix (HumanEvalPack, 6-language average)12.9% ✎▁·▁▁▄██▄█▁▅▅ beats 11/35medium · avg 56%▇▇ 5.5%sharp
math▁▃▃▅▆▅▄█▄▁▁▁ #388 of 398 * · 5.4% of its score · 1 test

Beats Code Llama 7B (base), Llama 2 13B Chat here, though they rank higher overall. Trails Qwen2 0.5B Instruct, Granite 3.1 1B-A400M Instruct, Gemma 2B (base) here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
GSM8K lab21.5% ✎▄▄▄·▄█·██·█▄ beats 3/12easy · avg 75%▇▇ 5.4%sharp · at its level

$ 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 StarCoderBase 15.5B · needs ~10.1 GB · from huggingface.co/mradermacher
    chat in the terminal
    llama cli -hf mradermacher/starcoderbase-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf mradermacher/starcoderbase-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 StarCoderBase 15.5B · needs ~10.1 GB · from huggingface.co/mradermacher

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/mradermacher/starcoderbase-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 StarCoderBase 15.5B · needs ~10.1 GB · from huggingface.co/mradermacher
    chat in the terminal
    ollama run hf.co/mradermacher/starcoderbase-GGUF:Q4_K_M

Other ways to run it

  1. Run it as published · needs ~63.28 GB · from huggingface.co/bigcode (official)
    uv run --with torch --with 'transformers<5' --with accelerate python -c "from transformers import pipeline; print(pipeline('text-generation', model='bigcode/starcoderbase', torch_dtype='auto', device_map='auto')('def print_hello_world():', max_new_tokens=40)[0]['generated_text'])"

Official weights and documentation: huggingface.co/bigcode ↗

$ how the score was worked out · StarCoder 2 and The Stack v2:… 48%, Granite Code Models: A Family… 45%, CRUXEval published evaluation… 6%

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
StarCoder 2 and The Stack v2: The Next Generation — authors' own runs · lab-reported4▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 48%CC BY-SA 4.0
Granite Code Models: A Family of Open Foundation Models for Code Intelligence — authors' own runs6▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 45%CC BY 4.0
CRUXEval published evaluation results (authors' generations and scores)2▇▇ 6%MIT
prior · pull toward the middle▇ 0%
weighted average= 33.0

$ 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