LMSteinshark / INFERENCE

Model playground

steinshark_369M_sft_enhanced_4

Stage: SFT

Current: Supervised fine-tuning. This 369M checkpoint completed three SFT epochs.

It is being evaluated for instruction following, fluent answers, and multi-turn continuity. Search and tool execution are not enabled.

Chat with the latest published checkpoint. Previous turns stay in the conversation, and you can adjust response length, temperature, and top-p.

Parameters

369.2M369,167,360 total

321,024 hidden width

2,048prompt + generated tokens

32,770steinshark_32k BPE

SFT ASSISTANT

Ask a question, then follow up naturally. Your messages are sent as conversation context for each response.

Response length Temperature 0.70 Top-p 0.90

Start a conversation with the model.