Quantization made by Richard Erkhov.

Github

Discord

Request more models

SELM-Llama-3-8B-Instruct-iter-3 - GGUF

• Model creator: https://huggingface.co/ZhangShenao/
• Original model: https://huggingface.co/ZhangShenao/SELM-Llama-3-8B-Instruct-iter-3/

Name	Quant method	Size
SELM-Llama-3-8B-Instruct-iter-3.Q2K.gguf	Q2K	2.96GB
SELM-Llama-3-8B-Instruct-iter-3.IQ3XS.gguf	IQ3XS	3.28GB
SELM-Llama-3-8B-Instruct-iter-3.IQ3S.gguf	IQ3S	3.43GB
SELM-Llama-3-8B-Instruct-iter-3.Q3KS.gguf	Q3K_S	3.41GB
SELM-Llama-3-8B-Instruct-iter-3.IQ3M.gguf	IQ3M	3.52GB
SELM-Llama-3-8B-Instruct-iter-3.Q3K.gguf	Q3K	3.74GB
SELM-Llama-3-8B-Instruct-iter-3.Q3KM.gguf	Q3K_M	3.74GB
SELM-Llama-3-8B-Instruct-iter-3.Q3KL.gguf	Q3K_L	4.03GB
SELM-Llama-3-8B-Instruct-iter-3.IQ4XS.gguf	IQ4XS	4.18GB
SELM-Llama-3-8B-Instruct-iter-3.Q40.gguf	Q40	4.34GB
SELM-Llama-3-8B-Instruct-iter-3.IQ4NL.gguf	IQ4NL	4.38GB
SELM-Llama-3-8B-Instruct-iter-3.Q4KS.gguf	Q4K_S	4.37GB
SELM-Llama-3-8B-Instruct-iter-3.Q4K.gguf	Q4K	4.58GB
SELM-Llama-3-8B-Instruct-iter-3.Q4KM.gguf	Q4K_M	4.58GB
SELM-Llama-3-8B-Instruct-iter-3.Q41.gguf	Q41	4.78GB
SELM-Llama-3-8B-Instruct-iter-3.Q50.gguf	Q50	5.21GB
SELM-Llama-3-8B-Instruct-iter-3.Q5KS.gguf	Q5K_S	5.21GB
SELM-Llama-3-8B-Instruct-iter-3.Q5K.gguf	Q5K	5.34GB
SELM-Llama-3-8B-Instruct-iter-3.Q5KM.gguf	Q5K_M	5.34GB
SELM-Llama-3-8B-Instruct-iter-3.Q51.gguf	Q51	5.65GB
SELM-Llama-3-8B-Instruct-iter-3.Q6K.gguf	Q6K	6.14GB
SELM-Llama-3-8B-Instruct-iter-3.Q80.gguf	Q80	7.95GB

Original model description:

---

license: mit
base_model: ZhangShenao/SELM-Llama-3-8B-Instruct-iter-2
tags:

• alignment-handbook
• dpo
• trl
• selm

datasets:

• HuggingFaceH4/ultrafeedback_binarized

model-index:

• name: SELM-Llama-3-8B-Instruct-iter-3

results: []

---

Self-Exploring Language Models: Active Preference Elicitation for Online Alignment.

SELM-Llama-3-8B-Instruct-iter-3

This model is a fine-tuned version of ZhangShenao/SELM-Llama-3-8B-Instruct-iter-2 using synthetic data based on on the HuggingFaceH4/ultrafeedbackbinarized dataset.

Model description

• Model type: A 8B parameter Llama3-instruct-based Self-Exploring Language Models (SELM).
• License: MIT

Results

	AlpacaEval 2.0 (LC WR)	MT-Bench (Average)
SELM-Llama-3-8B-Instruct-iter-3	33.47	8.29
SELM-Llama-3-8B-Instruct-iter-2	35.65	8.09
SELM-Llama-3-8B-Instruct-iter-1	32.02	7.92
Meta-Llama-3-8B-Instruct	24.31	7.93

Our model also ranks highly on WildBench! 🔥

Training hyperparameters

The following hyperparameters were used during training:

• alpha: 0.0001
• beta: 0.01
• trainbatchsize: 4
• seed: 42
• distributedtype: multi-GPU
• numdevices: 8
• gradientaccumulationsteps: 4
• totaltrainbatchsize: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• numepochs: 1

Framework versions

• Transformers 4.40.2
• Pytorch 2.1.2+cu121
• Datasets 2.14.6
• Tokenizers 0.19.1