CoolFace
Modelpublic

RichardErkhov/BreadAi_-_gpt-Youtube-gguf

sourceHugging Faceupdated 2y agoView on Hugging Face
0likes252downloads
Model Card

Quantization made by Richard Erkhov.

Github

Discord

Request more models

gpt-Youtube - GGUF

  • —Model creator: https://huggingface.co/BreadAi/
  • —Original model: https://huggingface.co/BreadAi/gpt-Youtube/
NameQuant methodSize
gpt-Youtube.Q2_K.ggufQ2_K0.07GB
gpt-Youtube.IQ3_XS.ggufIQ3_XS0.08GB
gpt-Youtube.IQ3_S.ggufIQ3_S0.08GB
gpt-Youtube.Q3_K_S.ggufQ3KS0.08GB
gpt-Youtube.IQ3_M.ggufIQ3_M0.08GB
gpt-Youtube.Q3_K.ggufQ3_K0.09GB
gpt-Youtube.Q3_K_M.ggufQ3KM0.09GB
gpt-Youtube.Q3_K_L.ggufQ3KL0.09GB
gpt-Youtube.IQ4_XS.ggufIQ4_XS0.09GB
gpt-Youtube.Q4_0.ggufQ4_00.1GB
gpt-Youtube.IQ4_NL.ggufIQ4_NL0.1GB
gpt-Youtube.Q4_K_S.ggufQ4KS0.1GB
gpt-Youtube.Q4_K.ggufQ4_K0.1GB
gpt-Youtube.Q4_K_M.ggufQ4KM0.1GB
gpt-Youtube.Q4_1.ggufQ4_10.1GB
gpt-Youtube.Q5_0.ggufQ5_00.11GB
gpt-Youtube.Q5_K_S.ggufQ5KS0.11GB
gpt-Youtube.Q5_K.ggufQ5_K0.12GB
gpt-Youtube.Q5_K_M.ggufQ5KM0.12GB
gpt-Youtube.Q5_1.ggufQ5_10.12GB
gpt-Youtube.Q6_K.ggufQ6_K0.13GB
gpt-Youtube.Q8_0.ggufQ8_00.16GB

Original model description: --- datasets:

  • —breadlicker45/youtube-comments-180k pipeline_tag: text-generation --- this is trained on 180K YouTube comments.

this is trained for 100k steps.

Open LLM Leaderboard Evaluation Results

Detailed results can be found here

MetricValue
Avg.24.86
ARC (25-shot)23.29
HellaSwag (10-shot)26.34
MMLU (5-shot)23.54
TruthfulQA (0-shot)48.63
Winogrande (5-shot)48.93
GSM8K (5-shot)0.0
DROP (3-shot)3.32