MSc_llama2_finetuned_model_secondData9

This model is a fine-tuned version of meta-llama/Llama-2-7b-chat-hf on the None dataset. It achieves the following results on the evaluation set:

• Loss: 0.7454

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

The following bitsandbytes quantization config was used during training:

• quant_method: bitsandbytes
• _load_in_8bit: False
• _load_in_4bit: True
• llm_int8_threshold: 6.0
• llm_int8_skip_modules: None
• llm_int8_enable_fp32_cpu_offload: False
• llm_int8_has_fp16_weight: False
• bnb_4bit_quant_type: nf4
• bnb_4bit_use_double_quant: True
• bnb_4bit_compute_dtype: bfloat16
• load_in_4bit: True
• load_in_8bit: False

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 3e-05
• train_batch_size: 32
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 64
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: cosine
• lr_scheduler_warmup_ratio: 0.03
• training_steps: 250

Training results

Training Loss	Epoch	Step	Validation Loss
4.0419	1.33	10	3.8603
3.6394	2.67	20	3.4241
3.165	4.0	30	2.8774
2.6005	5.33	40	2.3073
2.0686	6.67	50	1.8698
1.7422	8.0	60	1.6598
1.5451	9.33	70	1.4786
1.3602	10.67	80	1.2727
1.1005	12.0	90	0.9606
0.871	13.33	100	0.8730
0.8094	14.67	110	0.8396
0.7729	16.0	120	0.8150
0.7393	17.33	130	0.7961
0.7087	18.67	140	0.7818
0.6975	20.0	150	0.7702
0.6765	21.33	160	0.7626
0.6642	22.67	170	0.7570
0.6555	24.0	180	0.7529
0.6485	25.33	190	0.7503
0.6416	26.67	200	0.7474
0.6363	28.0	210	0.7464
0.6403	29.33	220	0.7458
0.6254	30.67	230	0.7455
0.6347	32.0	240	0.7451
0.6337	33.33	250	0.7454

Framework versions

• PEFT 0.4.0
• Transformers 4.38.2
• Pytorch 2.4.0+cu121
• Datasets 2.13.1
• Tokenizers 0.15.2