GLM-5.3 is live. Run Z.ai's latest model on Friendli Model APIs. Try it today ➜
Stee201
Available on FriendliAI
Run this model inference on single tenant GPU with unmatched speed and reliability at scale.
Run this model inference with full control and performance in your environment.
Model Details
Model Provider
Stee201
Model Tree
google/gemma-3-1b-it
Input Modalities
Output Modalities
Supported Functionality
This adapter is trained for retrieval-augmented use, and it will not behave correctly without the retrieved passages: the training examples always contain six passages in the system prompt, so the adapter must be prompted the same way. The system prompt has the form
markdown
Sei un assistente di finanza personale. L'utente ha conoscenze di finanza intermedie.RULES: ...DOCUMENT [passage-id]:<passage text>... six passages in total ...
followed by the user's question. The persona sentence states the reader's level, and it is what selects the register: the adapter was trained on three of them (basic, intermediate, advanced), so the same question returns an answer written at the requested level.
LoRA rank 8, alpha 16, dropout 0.05, on the attention and feed-forward projections, loss on answer tokens only. 4 epochs, learning rate 5e-5, checkpoint selected by lowest validation loss.