Modelpublic

C-H-Liu/bash2nl-qwen2.5-coder-7b-GGUF

sourceHugging Faceapache-2.0updated 6d agoView on Hugging Face

1likes66downloads

Model Card

bash2nl — Qwen2.5-Coder-7B QLoRA, GGUF q4KM

Explains a Bash command line in exactly one English sentence, phrased as an instruction starting with a verb. QLoRA fine-tune of Qwen/Qwen2.5-Coder-7B-Instruct, merged and quantized to q4KM.

find . -name "*.py"
-> Display the names of all files in the current directory and recursively into subdirectories with names that end in .py

Usage

bash

hf download C-H-Liu/bash2nl-qwen2.5-coder-7b-GGUF --local-dir bash2nl
cd bash2nl && ollama create bash2nl -f Modelfile
ollama run bash2nl 'ps -ef | grep nginx | awk "{print \$2}" | xargs kill -9'

The Modelfile pins the system prompt and this decoding:

option	value
`temperature`	0.0
`top_k`	1
`top_p`	1.0
`repeat_penalty`	1.0
`num_predict`	96
`num_ctx`	4096
stop	`<	im_end	>`,` <	endoftext	>`

Evaluation

920 held-out commands Judge columns are claude-sonnet-5 scoring a fixed 200-command subset.

metric	base	+ few-shot	this model
BLEU	14.59	17.67	33.49
chrF	44.15	45.82	52.21
ROUGE-L	39.09	40.64	53.24
judge acceptable+	0.880	0.900	0.890
judge wrong	0.120	0.100	0.110
avg words	18.0	17.2	12.4

License

Apache-2.0, matching the base model, whose weights this GGUF contains. Training data comes from the nl2bash project; consult it for the terms attached to that corpus.