Hippogriff 30B Chat GPTQ is an open-source language model by TheBloke. Features: 30b LLM, VRAM: 16.9GB, Context: 2K, License: other, Quantized, LLM Explorer Score: 0.07.
Can produce problematic outputs., Struggles with tasks related to math., May produce socially unacceptable text.
Considerations:
We recommend having a human curate or filter the outputs before releasing them, both to censor undesirable content and to improve the quality of the results.
Additional Notes
Build Hippogriff was built with Axolotl on 8xA100 80GB for 1.5 epochs taking approximately 12 hours.
Training Details
Data Sources:
OpenAssistant/oasst1 - cleaned dataset, similar to Guanaco, synthetic jokes generation and explanation derived from reddit jokes dataset, synthetic prose generation and rewriting self-chat, Q&A; based on provided context, self instruct augmented logic_inference_oa, de-duped pygmalion dataset, filtered down to RP data, cleaned, english only, 25%, riddle_sense - instruct augmented, hellaswag, updated for detailed explanations w 30K+ rows, gsm8k - instruct augmented, ewof/code-alpaca-instruct-unfiltered synthetic self chat dataset derived from about 1000 rows, subset of QingyiSi/Alpaca-CoT for roleplay and CoT, GPTeacher-General-Instruct, ARC-Easy & ARC-Challenge - instruct augmented for detailed responses, hellaswag - 5K row subset of instruct augmented for concise responses, metaeval/ScienceQA_text_only - instruct for concise responses, openai/summarize_from_feedback - instruct augmented tl;dr summarization
Training Time:
12 hours for 1.5 epochs
Hardware Used:
8xA100 80GB
Safety Evaluation
Ethical Considerations:
Hippogriff has not been aligned to human preferences with techniques like RLHF or deployed with in-the-loop filtering of responses. The model may produce problematic outputs.
Input Output
Input Format:
A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user's questions. USER: {prompt} ASSISTANT:
Performance Tips:
Never depend upon Hippogriff to produce factually accurate output.