t5-small-squad-qg-a2c-spt

This model is a fine-tuned version of lmqg/t5-small-squad-qg on the qg_squadshifts dataset. It achieves the following results on the evaluation set:

• Loss: 3.4424
• Bleu: 0.2369
• Precisions: [0.5087032407189018, 0.27403783600926385, 0.18636099825885083, 0.1320389623167492]
• Brevity Penalty: 0.9790
• Length Ratio: 0.9793
• Translation Length: 42398
• Reference Length: 43296

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: 0.0001
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 10
• label_smoothing_factor: 0.15

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Precisions	Brevity Penalty	Length Ratio	Translation Length	Reference Length
3.646	1.0	42	3.4501	0.2324	[0.5031057356491574, 0.26807773815061764, 0.18061915046796256, 0.12696709585121602]	0.9853	0.9854	42663	43296
3.5951	2.0	84	3.4456	0.2328	[0.5061518076850274, 0.27000913957435696, 0.1832138903455107, 0.12926178476134007]	0.9759	0.9762	42264	43296
3.5572	3.0	126	3.4427	0.2355	[0.505242954779515, 0.27049412978970455, 0.18334962341171734, 0.12953889087192133]	0.9867	0.9868	42724	43296
3.5295	4.0	168	3.4411	0.2351	[0.5057055646865461, 0.27130317702804174, 0.1838566316518527, 0.12948538278525568]	0.9836	0.9837	42590	43296
3.4945	5.0	210	3.4418	0.2359	[0.5068653913859875, 0.27228491562273527, 0.18446938010211442, 0.1297804417225878]	0.9839	0.9840	42605	43296
3.4771	6.0	252	3.4432	0.2375	[0.507522591245159, 0.2735272802567554, 0.18594051980269422, 0.13157208938693074]	0.9839	0.9840	42605	43296
3.46	7.0	294	3.4431	0.2377	[0.5092926294961487, 0.2746595987943041, 0.1869911632623497, 0.13212859294179272]	0.9803	0.9805	42453	43296
3.4656	8.0	336	3.4413	0.2368	[0.5082384555547698, 0.2738076663025953, 0.18616789908655937, 0.1317669419321012]	0.9796	0.9798	42423	43296
3.443	9.0	378	3.4425	0.2373	[0.5089378360532025, 0.27438532587485365, 0.18661869668658967, 0.13227530576778043]	0.9792	0.9794	42404	43296
3.4455	10.0	420	3.4424	0.2369	[0.5087032407189018, 0.27403783600926385, 0.18636099825885083, 0.1320389623167492]	0.9790	0.9793	42398	43296

Framework versions

• Transformers 4.27.4
• Pytorch 1.9.0
• Datasets 2.9.0
• Tokenizers 0.13.2