README.md

metadata

license: mit
tags:
  - generated_from_trainer
model-index:
  - name: rut5-base-absum-tech-support-calls
    results: []

rut5-base-absum-tech-support-calls

This model is a fine-tuned version of cointegrated/rut5-base-absum on an unknown dataset. It achieves the following results on the evaluation set:

• Loss: 1.4464
• Rouge-1: 0.5076
• Rouge-2: 0.3897
• Rouge-l: 0.4945
• Gen Len: 15.75
• Avg Rouge F: 0.4639

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: 3
• eval_batch_size: 1
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 100

Training results

Training Loss	Epoch	Step	Validation Loss	Rouge-1	Rouge-2	Rouge-l	Gen Len	Avg Rouge F
2.6017	2.78	50	2.0030	0.0	0.0	0.0	8.125	0.0
2.1413	5.56	100	1.5154	0.1125	0.0317	0.0958	11.5	0.08
1.6874	8.33	150	1.2364	0.3417	0.2312	0.325	13.25	0.2993
1.2272	11.11	200	1.1259	0.3605	0.2437	0.3291	14.25	0.3111
0.9384	13.89	250	1.0853	0.4505	0.3	0.4211	13.5	0.3905
0.7071	16.67	300	1.0607	0.3559	0.1368	0.3133	14.875	0.2687
0.5871	19.44	350	1.0346	0.5377	0.4194	0.5126	16.0	0.4899
0.4194	22.22	400	1.0672	0.5079	0.3819	0.4829	15.5	0.4576
0.3685	25.0	450	1.1284	0.5029	0.3835	0.4897	14.75	0.4587
0.2884	27.78	500	1.1729	0.5427	0.421	0.5164	15.875	0.4933
0.2368	30.56	550	1.1640	0.5326	0.421	0.5195	15.25	0.491
0.195	33.33	600	1.2053	0.5326	0.421	0.5195	15.25	0.491
0.1667	36.11	650	1.2525	0.4245	0.2717	0.4114	16.125	0.3692
0.1491	38.89	700	1.3346	0.5032	0.3897	0.4901	16.0	0.461
0.1122	41.67	750	1.3354	0.5094	0.4062	0.5094	15.375	0.475
0.1166	44.44	800	1.3685	0.5076	0.3897	0.4945	15.625	0.4639
0.0973	47.22	850	1.4157	0.5076	0.3897	0.4945	15.375	0.4639
0.0944	50.0	900	1.4523	0.5095	0.3897	0.4963	15.125	0.4652
0.0744	52.78	950	1.4221	0.5326	0.421	0.5195	15.25	0.491
0.0745	55.56	1000	1.4464	0.5076	0.3897	0.4945	15.75	0.4639

Framework versions

• Transformers 4.29.2
• Pytorch 2.0.1+cu118
• Tokenizers 0.13.3