OpenLeecher_-_llama_8b_lima_8_kto-gguf – AI Model by RichardErkhov | AlphaNeural AI

You can deploy this model and start earning money today!

RichardErkhov/
OpenLeecher_-_llama_8b_lima_8_kto-gguf

like

0

endpoints_compatible

Views

No views yet

Quantization made by Richard Erkhov.

Request more models

llama_8b_lima_8_kto - GGUF

Model creator: https://huggingface.co/OpenLeecher/
Original model: https://huggingface.co/OpenLeecher/llama_8b_lima_8_kto/

Name	Quant method	Size
llama_8b_lima_8_kto.Q2_K.gguf	Q2_K	2.96GB
llama_8b_lima_8_kto.IQ3_XS.gguf	IQ3_XS	3.28GB
llama_8b_lima_8_kto.IQ3_S.gguf	IQ3_S	3.43GB
llama_8b_lima_8_kto.Q3_K_S.gguf	Q3_K_S	3.41GB
llama_8b_lima_8_kto.IQ3_M.gguf	IQ3_M	3.52GB
llama_8b_lima_8_kto.Q3_K.gguf	Q3_K	3.74GB
llama_8b_lima_8_kto.Q3_K_M.gguf	Q3_K_M	3.74GB
llama_8b_lima_8_kto.Q3_K_L.gguf	Q3_K_L	4.03GB
llama_8b_lima_8_kto.IQ4_XS.gguf	IQ4_XS	4.18GB
llama_8b_lima_8_kto.Q4_0.gguf	Q4_0	4.34GB
llama_8b_lima_8_kto.IQ4_NL.gguf	IQ4_NL	4.38GB
llama_8b_lima_8_kto.Q4_K_S.gguf	Q4_K_S	4.37GB
llama_8b_lima_8_kto.Q4_K.gguf	Q4_K	4.58GB
llama_8b_lima_8_kto.Q4_K_M.gguf	Q4_K_M	4.58GB
llama_8b_lima_8_kto.Q4_1.gguf	Q4_1	4.78GB
llama_8b_lima_8_kto.Q5_0.gguf	Q5_0	5.21GB
llama_8b_lima_8_kto.Q5_K_S.gguf	Q5_K_S	5.21GB
llama_8b_lima_8_kto.Q5_K.gguf	Q5_K	5.34GB
llama_8b_lima_8_kto.Q5_K_M.gguf	Q5_K_M	5.34GB
llama_8b_lima_8_kto.Q5_1.gguf	Q5_1	5.65GB
llama_8b_lima_8_kto.Q6_K.gguf	Q6_K	6.14GB
llama_8b_lima_8_kto.Q8_0.gguf	Q8_0	7.95GB

Original model description:

library_name: transformers license: llama3 base_model: OpenLeecher/llama_8b_lima_8 tags:

trl
kto
llama-factory
generated_from_trainer model-index:
name: llama_8b_lima_8_kto results: []

llama_8b_lima_8_kto

This model is a fine-tuned version of OpenLeecher/llama_8b_lima_8 on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.1183
Rewards/chosen: 5.0516
Logps/chosen: -182.1999
Rewards/rejected: -5.2649
Logps/rejected: -221.3531
Rewards/margins: 10.3165
Kl: 9.8835

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: 1e-06
train_batch_size: 1
eval_batch_size: 1
seed: 66
distributed_type: multi-GPU
num_devices: 2
gradient_accumulation_steps: 12
total_train_batch_size: 24
total_eval_batch_size: 2
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: constant
lr_scheduler_warmup_steps: 25
num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss	Rewards/chosen	Logps/chosen	Rewards/rejected	Logps/rejected	Rewards/margins	Kl
0.3562	0.4995	157	0.2510	2.2688	-210.0280	-3.0690	-199.3941	5.3378	4.8987
0.3412	0.9989	314	0.1183	5.0516	-182.1999	-5.2649	-221.3531	10.3165	9.8835

Framework versions

Transformers 4.45.0
Pytorch 2.4.0
Datasets 2.21.0
Tokenizers 0.20.0