Modelpublic

RichardErkhov/abacaj_-_llama-161M-100B-gguf

sourceHugging Faceupdated 2y agoView on Hugging Face

0likes165downloads

Model Card

Quantization made by Richard Erkhov.

Request more models

llama-161M-100B - GGUF

—Model creator: https://huggingface.co/abacaj/
—Original model: https://huggingface.co/abacaj/llama-161M-100B/

Name	Quant method	Size
llama-161M-100B.Q2_K.gguf	Q2_K	0.1GB
llama-161M-100B.IQ3_XS.gguf	IQ3_XS	0.1GB
llama-161M-100B.IQ3_S.gguf	IQ3_S	0.1GB
llama-161M-100B.Q3_K_S.gguf	Q3KS	0.1GB
llama-161M-100B.IQ3_M.gguf	IQ3_M	0.1GB
llama-161M-100B.Q3_K.gguf	Q3_K	0.1GB
llama-161M-100B.Q3_K_M.gguf	Q3KM	0.1GB
llama-161M-100B.Q3_K_L.gguf	Q3KL	0.1GB
llama-161M-100B.IQ4_XS.gguf	IQ4_XS	0.1GB
llama-161M-100B.Q4_0.gguf	Q4_0	0.1GB
llama-161M-100B.IQ4_NL.gguf	IQ4_NL	0.1GB
llama-161M-100B.Q4_K_S.gguf	Q4KS	0.11GB
llama-161M-100B.Q4_K.gguf	Q4_K	0.11GB
llama-161M-100B.Q4_K_M.gguf	Q4KM	0.11GB
llama-161M-100B.Q4_1.gguf	Q4_1	0.11GB
llama-161M-100B.Q5_0.gguf	Q5_0	0.11GB
llama-161M-100B.Q5_K_S.gguf	Q5KS	0.12GB
llama-161M-100B.Q5_K.gguf	Q5_K	0.12GB
llama-161M-100B.Q5_K_M.gguf	Q5KM	0.12GB
llama-161M-100B.Q5_1.gguf	Q5_1	0.12GB
llama-161M-100B.Q6_K.gguf	Q6_K	0.15GB
llama-161M-100B.Q8_0.gguf	Q8_0	0.16GB

Original model description: --- library_name: transformers license: apache-2.0 ---

llama-161M

Trained on 100B tokens.

—1e-3 LR
—0.1 wd
—WSD scheduler with 10% decay
—80% code, 10% NL, 10% instruction data
—Dataset decontaminated against popular benchmarks following bigcode
—8x3090s 110~ hours

This is a base pretrained model and requires further fine tuning to be useful.

Model Details

[openai/openai_humaneval](https://huggingface.co/datasets/openai/openai_humaneval) (greedy)	[mbpp](https://huggingface.co/datasets/google-research-datasets/mbpp) (greedy)
9.2%	9.8%