| base_model: google/gemma-4-E4B-it | |
| base_model_revision: ee0ef6023621cff504d758262d4e04895a5af4a2 | |
| pipeline_tag: text-generation | |
| tags: [gemma4, ask, text-only, merged-lora, text-generation] | |
| private: true | |
| # ask gemma-4-E4B-it model | |
| Standalone text-only `Gemma4ForCausalLM` model created by merging the LoRA adapter from [`mizydorczyk/gemma-4-e4b-it-ask-lora`](https://huggingface.co/mizydorczyk/gemma-4-e4b-it-ask-lora) into the text tower of `google/gemma-4-E4B-it` at `ee0ef6023621cff504d758262d4e04895a5af4a2`. | |
| ## Sources | |
| | Source | Repository | Revision | | |
| |---|---|---| | |
| | Base model | [`google/gemma-4-E4B-it`](https://huggingface.co/google/gemma-4-E4B-it) | `ee0ef6023621cff504d758262d4e04895a5af4a2` | | |
| | LoRA adapter | [`mizydorczyk/gemma-4-e4b-it-ask-lora`](https://huggingface.co/mizydorczyk/gemma-4-e4b-it-ask-lora) | `5f8ad6dad0f1ac7931107ab8c3d0918f48477374` | | |
| ## Chat templates | |
| - `chat_template.jinja`: Google's canonical Gemma 4 inference template, used by the deployment endpoint. | |
| - `chat_template_training.jinja`: the reproducible training template; it renders identical prompts but adds TRL assistant-loss annotations. | |