julienp79/occitan-gemma-3-12b-it-rslora-sfttrainer

Add Q2_K GGUF quant for constrained hardware

Fix config.json: remove null quantization_config

Add merged model (bf16 shards), GGUF quants (gguf/), move adapter to adapter/

Update model weights and README (block size 768, lr 1e-4, max_grad_norm 1.0)

Upload README.md with huggingface_hub

Upload folder using huggingface_hub

Upload adapter_config.json with huggingface_hub

initial commit