Downloads · 30 days
96
1% of all-time downloads
helenk/gemma-4-E4B-finetune-GGUF
gemma-4-E4B-finetune-GGUF is a text generation model from helenk. Use it when you need the model to write or continue text. It is set up for gguf. The card lists the license as gemma.
Q4KM-quantized GGUF of the OfflineAid Stage-1 fine-tune. Drop-in for llama.cpp and Ollama.
Downloads · 30 days
96
1% of all-time downloads
All-time downloads
9.3K
Public
Repo size
16.9 GB
Likes
1
Public
Click a slice to open those files.
.gguf5.3 GB · 100%
From the Hugging Face model README
Q4_K_M-quantized GGUF of the OfflineAid Stage-1 fine-tune. Drop-in for llama.cpp and Ollama.
unsloth/gemma-4-E4B-ithelenk/gemma-4-E4B-lora — Unsloth fine-tune on Kaggle T4helenk/gemma-4-E4B-finetunepeft.merge_and_unload → llama.cpp/convert_hf_to_gguf.py → llama-quantize Q4_K_MHeld-out: 111 rows stratified per-language (37 EN + 37 ZH + 37 AR) from the 1,113-row helenkwok/offlineaid corpus, seed=3407. Both models served by Ollama (Q4_K_M). Greedy decoding, explicit "Answer in {language}" directive.
| Language | Metric | stock + RAG | ft + RAG | Δ |
|---|---|---|---|---|
| EN | ROUGE-L F1 | 0.688 | 0.699 | +0.011 |
| EN | Format-OK % | 91.9% | 94.6% | +2.7 pp |
| ZH | Format-OK % | 45.9% | 62.2% | +16.3 pp |
| AR | ROUGE-L F1 | 0.085 | 0.139 | +63% |
| AR | Format-OK % | 21.6% | 54.1% | +32.4 pp (2.5×) |
| all | ROUGE-L F1 | 0.334 | 0.355 | +0.021 |
| all | Format-OK % | 53.2% | 70.3% | +17.0 pp |
The fine-tune's value lives in multilingual robustness, especially Arabic (format-OK 21.6% → 54.1%, ROUGE-L +63%). Reproducible via bash scripts/tier_a_pipeline_eval_only.sh in the OfflineAid repo.
ollama pull hf.co/helenk/gemma-4-E4B-finetune-GGUF
Or via a local Modelfile:
FROM /path/to/gemma-4-E4B-offlineaid-Q4_K_M.gguf
RENDERER gemma4
PARSER gemma4
PARAMETER num_ctx 32768
PARAMETER stop "<turn|>"
PARAMETER temperature 0.0
ollama create offlineaid-e4b -f Modelfile
ollama run offlineaid-e4b
./llama-cli \
-m gemma-4-E4B-offlineaid-Q4_K_M.gguf \
-p "Answer in Simplified Chinese.\n\nQUESTION: ..." \
--temp 0.0 -n 256
Stage 3 of the OfflineAid pipeline — Mac-side pack-builder agent loop. Pixel 7 production deployment uses stock Gemma 4 E2B + retrieval, not this fine-tune; see the project writeup for the architectural rationale.
Inherits Google's Gemma Terms of Use. Training data (helenkwok/offlineaid) is CC-BY-4.0.
helenk/gemma-4-E4B-finetunehelenk/gemma-4-E4B-lorahelenk/gemma-4-E2B-finetune-GGUF