RichardErkhov/mtzig_-_debug_full_test-gguf

Quantization made by Richard Erkhov.

Github

Discord

Request more models

debug_full_test - GGUF

• Model creator: https://huggingface.co/mtzig/
• Original model: https://huggingface.co/mtzig/debug_full_test/

Name	Quant method	Size
debug_full_test.Q2_K.gguf	Q2_K	0.07GB
debug_full_test.IQ3_XS.gguf	IQ3_XS	0.08GB
debug_full_test.IQ3_S.gguf	IQ3_S	0.08GB
debug_full_test.Q3_K_S.gguf	Q3_K_S	0.08GB
debug_full_test.IQ3_M.gguf	IQ3_M	0.08GB
debug_full_test.Q3_K.gguf	Q3_K	0.09GB
debug_full_test.Q3_K_M.gguf	Q3_K_M	0.09GB
debug_full_test.Q3_K_L.gguf	Q3_K_L	0.09GB
debug_full_test.IQ4_XS.gguf	IQ4_XS	0.09GB
debug_full_test.Q4_0.gguf	Q4_0	0.1GB
debug_full_test.IQ4_NL.gguf	IQ4_NL	0.1GB
debug_full_test.Q4_K_S.gguf	Q4_K_S	0.1GB
debug_full_test.Q4_K.gguf	Q4_K	0.1GB
debug_full_test.Q4_K_M.gguf	Q4_K_M	0.1GB
debug_full_test.Q4_1.gguf	Q4_1	0.1GB
debug_full_test.Q5_0.gguf	Q5_0	0.11GB
debug_full_test.Q5_K_S.gguf	Q5_K_S	0.11GB
debug_full_test.Q5_K.gguf	Q5_K	0.12GB
debug_full_test.Q5_K_M.gguf	Q5_K_M	0.12GB
debug_full_test.Q5_1.gguf	Q5_1	0.12GB
debug_full_test.Q6_K.gguf	Q6_K	0.13GB
debug_full_test.Q8_0.gguf	Q8_0	0.17GB

Original model description:

library_name: transformers base_model: TinyPixel/small-llama2 tags: - generated_from_trainer metrics: - accuracy - precision - recall - f1 model-index: - name: debug_full_test results: []

debug_full_test

This model is a fine-tuned version of TinyPixel/small-llama2 on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.2665
• Accuracy: 0.6300
• Precision: 0.375
• Recall: 0.0299
• F1: 0.0553

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: 2e-05
• train_batch_size: 8
• eval_batch_size: 8
• seed: 42
• distributed_type: multi-GPU
• num_devices: 4
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• total_eval_batch_size: 32
• 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: cosine
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy	Precision	Recall	F1
1.6327	1.0	5	1.2665	0.6300	0.375	0.0299	0.0553

Framework versions

• Transformers 4.46.0
• Pytorch 2.5.1+cu124
• Datasets 3.1.0
• Tokenizers 0.20.3