You can deploy this model and start earning money today!

RichardErkhov/
hxssgaa_-_llama-3-8b-dpo-full-gguf

Name: hxssgaa_-_llama-3-8b-dpo-full-gguf
Availability: InStock

Views

No views yet

Quantization made by Richard Erkhov.

Github

Discord

Request more models

llama-3-8b-dpo-full - GGUF

Model creator: https://huggingface.co/hxssgaa/
Original model: https://huggingface.co/hxssgaa/llama-3-8b-dpo-full/

Name	Quant method	Size
llama-3-8b-dpo-full.Q2_K.gguf	Q2_K	2.96GB
llama-3-8b-dpo-full.IQ3_XS.gguf	IQ3_XS	3.28GB
llama-3-8b-dpo-full.IQ3_S.gguf	IQ3_S	3.43GB
llama-3-8b-dpo-full.Q3_K_S.gguf	Q3_K_S	3.41GB
llama-3-8b-dpo-full.IQ3_M.gguf	IQ3_M	3.52GB
llama-3-8b-dpo-full.Q3_K.gguf	Q3_K	3.74GB
llama-3-8b-dpo-full.Q3_K_M.gguf	Q3_K_M	3.74GB
llama-3-8b-dpo-full.Q3_K_L.gguf	Q3_K_L	4.03GB
llama-3-8b-dpo-full.IQ4_XS.gguf	IQ4_XS	4.18GB
llama-3-8b-dpo-full.Q4_0.gguf	Q4_0	4.34GB
llama-3-8b-dpo-full.IQ4_NL.gguf	IQ4_NL	4.38GB
llama-3-8b-dpo-full.Q4_K_S.gguf	Q4_K_S	4.37GB
llama-3-8b-dpo-full.Q4_K.gguf	Q4_K	4.58GB
llama-3-8b-dpo-full.Q4_K_M.gguf	Q4_K_M	4.58GB
llama-3-8b-dpo-full.Q4_1.gguf	Q4_1	4.78GB
llama-3-8b-dpo-full.Q5_0.gguf	Q5_0	5.21GB
llama-3-8b-dpo-full.Q5_K_S.gguf	Q5_K_S	5.21GB
llama-3-8b-dpo-full.Q5_K.gguf	Q5_K	5.34GB
llama-3-8b-dpo-full.Q5_K_M.gguf	Q5_K_M	5.34GB
llama-3-8b-dpo-full.Q5_1.gguf	Q5_1	5.65GB
llama-3-8b-dpo-full.Q6_K.gguf	Q6_K	6.14GB
llama-3-8b-dpo-full.Q8_0.gguf	Q8_0	7.95GB

Original model description:

library_name: transformers license: llama3 base_model: meta-llama/Meta-Llama-3-8B-Instruct tags:

alignment-handbook
trl
dpo
generated_from_trainer
trl
dpo
generated_from_trainer datasets:
trl-lib/ultrafeedback_binarized model-index:
name: llama-3-8b-dpo-full results: []

llama-3-8b-dpo-full

This model is a fine-tuned version of meta-llama/Meta-Llama-3-8B-Instruct on the trl-lib/ultrafeedback_binarized dataset. It achieves the following results on the evaluation set:

Loss: 0.6491
Rewards/chosen: -0.1814
Rewards/rejected: -0.2255
Rewards/accuracies: 0.5625
Rewards/margins: 0.0441
Logps/rejected: -419.1795
Logps/chosen: -335.9990
Logits/rejected: -1.1373
Logits/chosen: -1.0280

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

learning_rate: 3e-07
train_batch_size: 2
eval_batch_size: 4
seed: 42
distributed_type: multi-GPU
num_devices: 32
gradient_accumulation_steps: 8
total_train_batch_size: 512
total_eval_batch_size: 128
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.1
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Rewards/rejected	Rewards/accuracies	Rewards/margins	Logps/rejected	Logps/chosen	Logits/rejected	Logits/chosen
0.6411	0.8239	100	0.6494	-0.1752	-0.2195	0.5625	0.0443	-418.5782	-335.3811	-1.1582	-1.0463

Framework versions

Transformers 4.45.1
Pytorch 2.4.0+cu121
Datasets 3.0.0
Tokenizers 0.20.0

RichardErkhov/hxssgaa_-_llama-3-8b-dpo-full-gguf

Original model description:

llama-3-8b-dpo-full

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

RichardErkhov/
hxssgaa_-_llama-3-8b-dpo-full-gguf