Quantization made by Richard Erkhov.

llama3-1_8b_oh_v3.1_wo_collective_cognition - GGUF

Model creator: https://huggingface.co/mlfoundations-dev/
Original model: https://huggingface.co/mlfoundations-dev/llama3-1_8b_oh_v3.1_wo_collective_cognition/

Name	Quant method	Size
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q2_K.gguf	Q2_K	2.96GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.IQ3_XS.gguf	IQ3_XS	3.28GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.IQ3_S.gguf	IQ3_S	3.43GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q3_K_S.gguf	Q3_K_S	3.41GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.IQ3_M.gguf	IQ3_M	3.52GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q3_K.gguf	Q3_K	3.74GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q3_K_M.gguf	Q3_K_M	3.74GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q3_K_L.gguf	Q3_K_L	4.03GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.IQ4_XS.gguf	IQ4_XS	4.18GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q4_0.gguf	Q4_0	4.34GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.IQ4_NL.gguf	IQ4_NL	4.38GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q4_K_S.gguf	Q4_K_S	4.37GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q4_K.gguf	Q4_K	4.58GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q4_K_M.gguf	Q4_K_M	4.58GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q4_1.gguf	Q4_1	4.78GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q5_0.gguf	Q5_0	5.21GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q5_K_S.gguf	Q5_K_S	5.21GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q5_K.gguf	Q5_K	5.34GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q5_K_M.gguf	Q5_K_M	5.34GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q5_1.gguf	Q5_1	5.65GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q6_K.gguf	Q6_K	6.14GB
llama3-1_8b_oh_v3.1_wo_collective_cognition.Q8_0.gguf	Q8_0	7.95GB

Original model description:

library_name: transformers license: llama3.1 base_model: meta-llama/Meta-Llama-3.1-8B tags: - llama-factory - full - generated_from_trainer model-index: - name: llama3-1_8b_oh_v3.1_wo_collective_cognition results: []

llama3-1_8b_oh_v3.1_wo_collective_cognition

This model is a fine-tuned version of meta-llama/Meta-Llama-3.1-8B on the mlfoundations-dev/oh_v3.1_wo_collective_cognition dataset. It achieves the following results on the evaluation set:

Loss: 0.6409

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: 5e-06
train_batch_size: 16
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 32
total_train_batch_size: 512
total_eval_batch_size: 256
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: constant
lr_scheduler_warmup_ratio: 0.1
lr_scheduler_warmup_steps: 1738
num_epochs: 3.0

Training results

Training Loss	Epoch	Step	Validation Loss
0.6503	1.0	423	0.6491
0.602	2.0	846	0.6394
0.5681	3.0	1269	0.6409

Framework versions

Transformers 4.46.1
Pytorch 2.4.0
Datasets 3.0.2
Tokenizers 0.20.3