MSc_llama2_finetuned_model_updatePara

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: 0.4329

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: 0.0001
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 4
• 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: 0.1
• training_steps: 400

Training results

Training Loss	Epoch	Step	Validation Loss
16.5659	1.21	10	1.7444
1.1543	2.42	20	0.8871
0.7821	3.64	30	0.6674
0.6295	4.85	40	0.5773
0.5642	6.06	50	0.5303
0.5251	7.27	60	0.4983
0.4969	8.48	70	0.4813
0.4805	9.7	80	0.4689
0.4702	10.91	90	0.4585
0.4651	12.12	100	0.4512
0.4472	13.33	110	0.4462
0.4466	14.55	120	0.4415
0.4413	15.76	130	0.4385
0.4398	16.97	140	0.4398
0.4314	18.18	150	0.4382
0.4313	19.39	160	0.4313
0.4314	20.61	170	0.4294
0.4189	21.82	180	0.4279
0.4203	23.03	190	0.4285
0.4164	24.24	200	0.4279
0.4183	25.45	210	0.4268
0.4107	26.67	220	0.4268
0.4068	27.88	230	0.4252
0.4082	29.09	240	0.4266
0.4074	30.3	250	0.4299
0.4025	31.52	260	0.4266
0.4038	32.73	270	0.4264
0.4008	33.94	280	0.4284
0.4002	35.15	290	0.4279
0.3937	36.36	300	0.4307
0.4033	37.58	310	0.4299
0.3974	38.79	320	0.4307
0.3951	40.0	330	0.4312
0.3948	41.21	340	0.4327
0.3971	42.42	350	0.4321
0.3958	43.64	360	0.4327
0.3906	44.85	370	0.4333
0.3977	46.06	380	0.4328
0.3956	47.27	390	0.4330
0.3956	48.48	400	0.4329

Framework versions

• PEFT 0.4.0
• Transformers 4.38.2
• Pytorch 2.3.0+cu121
• Datasets 2.13.1
• Tokenizers 0.15.2