RichardErkhov/DongfuJiang_-_prm_math_only_hf-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

prm_math_only_hf - GGUF

• Model creator: https://huggingface.co/DongfuJiang/
• Original model: https://huggingface.co/DongfuJiang/prm_math_only_hf/

Name	Quant method	Size
prm_math_only_hf.Q2_K.gguf	Q2_K	2.96GB
prm_math_only_hf.IQ3_XS.gguf	IQ3_XS	3.28GB
prm_math_only_hf.IQ3_S.gguf	IQ3_S	3.43GB
prm_math_only_hf.Q3_K_S.gguf	Q3_K_S	3.41GB
prm_math_only_hf.IQ3_M.gguf	IQ3_M	3.52GB
prm_math_only_hf.Q3_K.gguf	Q3_K	3.74GB
prm_math_only_hf.Q3_K_M.gguf	Q3_K_M	3.74GB
prm_math_only_hf.Q3_K_L.gguf	Q3_K_L	4.03GB
prm_math_only_hf.IQ4_XS.gguf	IQ4_XS	4.18GB
prm_math_only_hf.Q4_0.gguf	Q4_0	4.34GB
prm_math_only_hf.IQ4_NL.gguf	IQ4_NL	4.38GB
prm_math_only_hf.Q4_K_S.gguf	Q4_K_S	4.37GB
prm_math_only_hf.Q4_K.gguf	Q4_K	4.58GB
prm_math_only_hf.Q4_K_M.gguf	Q4_K_M	4.58GB
prm_math_only_hf.Q4_1.gguf	Q4_1	4.78GB
prm_math_only_hf.Q5_0.gguf	Q5_0	5.21GB
prm_math_only_hf.Q5_K_S.gguf	Q5_K_S	5.21GB
prm_math_only_hf.Q5_K.gguf	Q5_K	5.34GB
prm_math_only_hf.Q5_K_M.gguf	Q5_K_M	5.34GB
prm_math_only_hf.Q5_1.gguf	Q5_1	5.65GB
prm_math_only_hf.Q6_K.gguf	Q6_K	6.14GB
prm_math_only_hf.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-Instruct tags: - llama-factory - full - generated_from_trainer model-index: - name: prm_version3_full_hf results: []

prm_version3_full_hf

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

• Loss: 0.1149

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: 1
• eval_batch_size: 1
• seed: 42
• distributed_type: multi-GPU
• num_devices: 8
• gradient_accumulation_steps: 8
• total_train_batch_size: 64
• total_eval_batch_size: 8
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.05
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss
0.1391	0.3935	500	0.1477
0.1183	0.7871	1000	0.1185

Framework versions

• Transformers 4.45.0
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3