llama-fintech-13

This model is a fine-tuned version of decapoda-research/llama-7b-hf on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.1889

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: QuantizationMethod.BITS_AND_BYTES
• 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: float32

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 2e-05
• train_batch_size: 4
• eval_batch_size: 8
• seed: 1234
• gradient_accumulation_steps: 32
• total_train_batch_size: 128
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 100
• num_epochs: 5

Training results

Training Loss	Epoch	Step	Validation Loss
2.4519	0.15	50	2.4590
2.2828	0.3	100	2.1885
1.9233	0.45	150	1.5158
1.6426	0.6	200	1.2491
1.6084	0.76	250	1.1512
1.5972	0.91	300	1.1561
1.5723	1.06	350	1.1782
1.5798	1.21	400	1.1833
1.566	1.36	450	1.1831
1.5625	1.51	500	1.1870
1.5514	1.66	550	1.1839
1.5381	1.81	600	1.1758
1.5518	1.96	650	1.1780
1.5321	2.12	700	1.1735
1.5406	2.27	750	1.1768
1.5275	2.42	800	1.1784
1.543	2.57	850	1.1788
1.5292	2.72	900	1.1851
1.5471	2.87	950	1.1820
1.509	3.02	1000	1.1858
1.5281	3.17	1050	1.1778
1.5146	3.33	1100	1.1835
1.5086	3.48	1150	1.1843
1.5379	3.63	1200	1.1854
1.5295	3.78	1250	1.1835
1.5185	3.93	1300	1.1871
1.5261	4.08	1350	1.1863
1.5215	4.23	1400	1.1873
1.5209	4.38	1450	1.1900
1.518	4.53	1500	1.1891
1.5287	4.69	1550	1.1887
1.5141	4.84	1600	1.1875
1.5263	4.99	1650	1.1889

Framework versions

• PEFT 0.6.0.dev0
• Transformers 4.32.1
• Pytorch 2.0.1
• Datasets 2.14.4
• Tokenizers 0.13.3