Made using mergekit-gui.

Unapologetically ~~ripping off~~ taking inspiration from ZeroXClem/Qwen3-4B-ChromaticCoder and prithivMLmods excellent Qwen3 code tunes.

prithivMLmods/Blitzar-Coder-4B-F.1 is a strong coder with extra tuning for rust (and 10+ others).
prithivMLmods/Bootes-Qwen3_Coder-Reasoning is a strong coder that works well on rust and has a tarser "personality" that I found got to the solution quickly.
prithivMLmods/Lacaille-MoT-4B-Supreme2 as a base provides a strong foundation for coding ("thinking", math, structured outputs etc.).

The intention is for this to be a solid rust coder + strong programming generalist.

Quantizations courtesy of team mradermacher

GGUF: GGUF
iMatrix: iMatrix-GGUF

name: Qwen3-4B-PastelCoder
base_model: prithivMLmods/Lacaille-MoT-4B-Supreme2
dtype: bfloat16
merge_method: model_stock
models:
  - model: prithivMLmods/Blitzar-Coder-4B-F.1
  - model: prithivMLmods/Bootes-Qwen3_Coder-Reasoning
tokenizer_source: prithivMLmods/Lacaille-MoT-4B-Supreme2

Downloads last month: 52

Safetensors

Model size

4B params

Tensor type

BF16

Inference Providers NEW

This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for sillykiwi/Qwen3-4B-PastelCoder

prithivMLmods/Blitzar-Coder-4B-F.1

prithivMLmods/Bootes-Qwen3_Coder-Reasoning

prithivMLmods/Lacaille-MoT-4B-Supreme2

Merge model

this model

Quantizations

2 models