julienp79/ occitan-gemma-3-4b-it-dora-sfttrainer:Q8_0

1 yesterday

Gemma 3 4B Instruct fine-tuned on Occitan via DoRA (r=16). Lightweight model for resource-constrained setups. Q2_K, Q4_K_M, Q5_K_M, Q8_0, f16. 

ollama run julienp79/occitan-gemma-3-4b-it-dora-sfttrainer:Q8_0

Details

yesterday

7b8ecb2f90f1 · 4.1GB ·

model
gemma3
·
3.88B
·
Q8_0
template
{{- $systemPromptAdded := false }} {{- range $i, $_ := .Messages }} {{- $last := eq (len (slice $.Me
system
Ès un escritor e grammarista occitan lengadocian. Respon unicament en occitan lengadocian. Escriu d
params
{ "num_ctx": 768, "stop": [ "<end_of_turn>" ], "temperature": 0.7, "top_

Readme

Occitan Lengadocian — Gemma 3 4B DoRA

Fine-tune of Gemma 3 4B Instruct on Occitan in the Lengadocian dialect (IEO grafia classica norm) using DoRA (weight-decomposed low-rank adaptation).

Specialist model for regional literary vocabulary. Produces the densest Hérault-specific material culture vocabulary of any 4B model in the collection — strong on landscape, rural life, and local colour.

Strengths

Distinctive vocabulary not found in other models: sornarutava, fenialha, cencha, cledisses miralhadas, fustatge escrincelat, atrivavan los graissals liscs, esparlongat, gorgolinar. Best choice when regional specificity and literary texture matter more than structured task performance.

Limitations

Weaker than the RS-LoRA model on journalistic register and grammatical metalanguage tasks.

System prompt

Ès un escritor e grammarista occitan lengadocian. Respon unicament en
occitan lengadocian. Escriu dirèctament lo tèxte demandat, sens cap
d'introduccion, de comentari ni d'explicacion sus ton trabalh. Pas de
preamble. Pas de version multiplas. Pas de traduccion.

Recommended quantisation

Quant Size Use case
Q4_K_M ~2.5 GB Recommended
Q5_K_M ~3.0 GB Slightly better quality
Q8_0 ~4.5 GB Near-lossless
Q2_K ~1.6 GB Minimal RAM

Training

DoRA · r=16 · α=32 · block_size=1024 · SFTTrainer
RTX 3060 12GB · ~5h30m