smol_llama-220M-GQA-32k-theta-sft-limarp

Experimental model meant to serve as a long-context speculative decoding model. This one is specifically for models trained on the LimaRP prompt format.

Created using Doctor-Shotgun/smol_llama-220M-GQA-32k-theta-sft and finetuning at 32768 context length on the LimaRP dataset.

This variant uses the rope theta (rope frequency base) method for context extension.

The trained instruction format is LimaRP Alpaca:

### Instruction:
Character's Persona: {bot character description}

User's Persona: {user character description}

Scenario: {what happens in the story}

Play the role of Character. Taking the above information into consideration, you must engage in a roleplaying chat with User below this line. Do not write dialogues and narration for User.

### Input:
User: {utterance}

### Response:
Character: {utterance}

### Input
User: {utterance}

### Response:
Character: {utterance}

(etc.)