Add model card with training recipe + sibling links

README.md

@@ -0,0 +1,79 @@
+---
+base_model: unsloth/gemma-4-E2B-it
+library_name: transformers
+license: gemma
+pipeline_tag: text-generation
+language:
+  - en
+  - zh
+  - ar
+tags:
+  - gemma
+  - gemma-4
+  - unsloth
+  - peft
+  - lora
+  - fine-tuned
+  - merged
+  - rag
+  - offlineaid
+  - australian-consumer-safety
+  - anti-scam
+  - disaster-response
+datasets:
+  - helenkwok/offlineaid
+---
+
+# OfflineAid — Gemma 4 E2B fine-tune (merged fp16 safetensors)
+
+Stage-1 fine-tune of [`unsloth/gemma-4-E2B-it`](https://huggingface.co/unsloth/gemma-4-E2B-it) for the [OfflineAid](https://github.com/helenkwok/offlineaid) project — an offline AI assistant for Australian consumer-safety scenarios (anti-scam, disaster, travel) targeting the [Kaggle Gemma 4 Good Hackathon](https://www.kaggle.com/competitions/gemma-4-good-hackathon).
+
+This repo contains the **fully merged fp16 safetensors** (3 shards, ~9.5 GB) produced by `peft.PeftModel.merge_and_unload()` from the LoRA adapter at [`helenk/gemma-4-E2B-lora`](https://huggingface.co/helenk/gemma-4-E2B-lora). For the Q4_K_M GGUF quantization, see [`helenk/gemma-4-E2B-finetune-GGUF`](https://huggingface.co/helenk/gemma-4-E2B-finetune-GGUF).
+
+The E2B variant is the smaller sibling of [`helenk/gemma-4-E4B-finetune`](https://huggingface.co/helenk/gemma-4-E4B-finetune). It targets lower-memory inference (Pixel 7 class) and was trained on the same corpus with identical hyperparameters via [`scripts/render_finetune_variant.py`](https://github.com/helenkwok/offlineaid/blob/master/repositories/offlineaid-pack-builder/scripts/render_finetune_variant.py).
+
+## Training
+
+- **Method:** Unsloth LoRA on Kaggle T4
+- **Adapter source:** [`helenk/gemma-4-E2B-lora`](https://huggingface.co/helenk/gemma-4-E2B-lora)
+- **Base:** [`unsloth/gemma-4-E2B-it`](https://huggingface.co/unsloth/gemma-4-E2B-it)
+- **LoRA config:** r=16, α=16, dropout=0, target = `q,k,v,o,gate,up,down_proj`, vision layers off
+- **Chat template:** `gemma-4-thinking`
+- **Loss:** `train_on_responses_only` (mask user + evidence turn, train only on answer tokens)
+- **Data:** [`helenkwok/offlineaid`](https://huggingface.co/datasets/helenkwok/offlineaid) v3 — 1,002-row train split (90/10 stratified per-language EN/ZH/AR from 1,113-row total, seed=3407). Each row is `{instruction, input (= verbatim evidence_quote from .gov.au source), output (grounded answer in target language), language}`.
+- **Hyperparams:** 2 epochs, batch 8 (per_device 2 × grad_accum 4), lr 2e-5, warmup 5, fp16, weight decay 0.01, seed 3407
+- **Notebook:** Identical canonical notebook as E4B with a single `MODEL_VARIANT = "E2B"` switch.
+
+## Merge recipe
+
+```python
+from peft import PeftModel
+from transformers import AutoModelForCausalLM, AutoTokenizer
+
+base = AutoModelForCausalLM.from_pretrained("unsloth/gemma-4-E2B-it", torch_dtype="float16", device_map="cpu")
+# Unsloth wraps Linear layers in Gemma4ClippableLinear — strip before peft injection
+# (see scripts/merge_e2b_lora.py for the unwrap helper)
+model = PeftModel.from_pretrained(base, "helenk/gemma-4-E2B-lora").merge_and_unload()
+model.save_pretrained("gemma-4-E2B-offlineaid-merged", safe_serialization=True, max_shard_size="5GB")
+AutoTokenizer.from_pretrained("helenk/gemma-4-E2B-lora").save_pretrained("gemma-4-E2B-offlineaid-merged")
+```
+
+Full script: [`scripts/merge_e2b_lora.py`](https://github.com/helenkwok/offlineaid/blob/master/repositories/offlineaid-pack-builder/scripts/merge_e2b_lora.py).
+
+## Tier A held-out eval
+
+The Tier A eval was run against the E4B variant (writeup-grade evidence pillar). E2B numbers are not separately reported here; see [`helenk/gemma-4-E4B-finetune-GGUF`](https://huggingface.co/helenk/gemma-4-E4B-finetune-GGUF) for the methodology and results table. Both variants share identical training data, recipe, and merge pipeline.
+
+## Intended use
+
+Lower-memory variant of the OfflineAid Stage-1 fine-tune. Pixel 7 production deployment uses *stock* `gemma-4-E2B-it.litertlm` plus retrieval, not this fine-tune; see [the project writeup](https://github.com/helenkwok/offlineaid) for the architectural rationale.
+
+## License
+
+Inherits Google's [Gemma Terms of Use](https://ai.google.dev/gemma/terms). Training data ([`helenkwok/offlineaid`](https://huggingface.co/datasets/helenkwok/offlineaid)) is CC-BY-4.0.
+
+## Sibling repos
+
+- LoRA adapter (~30 MB): [`helenk/gemma-4-E2B-lora`](https://huggingface.co/helenk/gemma-4-E2B-lora)
+- Q4_K_M GGUF for Ollama (~3.4 GB): [`helenk/gemma-4-E2B-finetune-GGUF`](https://huggingface.co/helenk/gemma-4-E2B-finetune-GGUF)
+- Larger E4B variant: [`helenk/gemma-4-E4B-finetune`](https://huggingface.co/helenk/gemma-4-E4B-finetune)