Tounsify-v0.7

This model is a fine-tuned version of Helsinki-NLP/opus-mt-en-ar on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.4410
• Bleu: 24.686
• Gen Len: 7.1333

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: 16
• eval_batch_size: 16
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• num_epochs: 100
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Bleu	Gen Len
No log	1.0	8	2.9143	16.1913	7.4
No log	2.0	16	2.6238	17.5534	6.9333
No log	3.0	24	2.4215	11.3684	6.3333
No log	4.0	32	2.2967	12.1601	6.3
No log	5.0	40	2.2459	12.4837	7.0333
No log	6.0	48	2.2333	13.2937	6.5667
No log	7.0	56	2.2264	18.1441	6.7667
No log	8.0	64	2.2167	14.5825	6.5333
No log	9.0	72	2.2064	15.1734	6.6667
No log	10.0	80	2.1951	14.6563	7.0333
No log	11.0	88	2.2060	19.0714	6.6333
No log	12.0	96	2.2088	21.5449	6.5333
No log	13.0	104	2.2517	21.4297	6.5333
No log	14.0	112	2.2584	24.6131	6.6
No log	15.0	120	2.2411	24.7358	6.6667
No log	16.0	128	2.2464	24.7358	6.6667
No log	17.0	136	2.2502	24.7358	6.6667
No log	18.0	144	2.2567	24.7358	6.6667
No log	19.0	152	2.2496	24.7358	6.6667
No log	20.0	160	2.2511	24.5996	6.8
No log	21.0	168	2.2668	24.5996	6.8
No log	22.0	176	2.2805	24.7358	6.6667
No log	23.0	184	2.2875	24.7358	6.6667
No log	24.0	192	2.2900	24.7358	6.6667
No log	25.0	200	2.2828	21.51	6.6667
No log	26.0	208	2.2676	21.51	6.6667
No log	27.0	216	2.2684	24.7358	6.6667
No log	28.0	224	2.2725	24.7358	6.6667
No log	29.0	232	2.2768	24.7358	6.6667
No log	30.0	240	2.2810	24.7358	6.6667
No log	31.0	248	2.2958	24.7358	6.6667
No log	32.0	256	2.3036	24.7358	6.6667
No log	33.0	264	2.3120	24.7358	6.7333
No log	34.0	272	2.3205	24.7358	6.7333
No log	35.0	280	2.3305	24.7358	6.7333
No log	36.0	288	2.3413	24.9721	6.7333
No log	37.0	296	2.3424	24.7358	6.7333
No log	38.0	304	2.3472	24.7358	6.7333
No log	39.0	312	2.3526	24.7358	6.7333
No log	40.0	320	2.3579	24.7358	6.7333
No log	41.0	328	2.3630	24.7358	6.6667
No log	42.0	336	2.3628	24.7358	6.6667
No log	43.0	344	2.3637	24.8163	6.7667
No log	44.0	352	2.3619	24.8163	6.7667
No log	45.0	360	2.3584	24.8163	6.7667
No log	46.0	368	2.3562	24.8163	6.7667
No log	47.0	376	2.3605	24.8163	6.7667
No log	48.0	384	2.3680	24.8163	6.8333
No log	49.0	392	2.3774	24.686	6.9667
No log	50.0	400	2.3819	24.686	6.9667
No log	51.0	408	2.3850	24.686	6.9667
No log	52.0	416	2.3902	24.686	6.9667
No log	53.0	424	2.3935	24.686	6.9667
No log	54.0	432	2.3969	24.686	6.9667
No log	55.0	440	2.3988	24.686	6.9667
No log	56.0	448	2.3992	24.686	6.9667
No log	57.0	456	2.3986	24.686	6.9667
No log	58.0	464	2.3983	24.686	6.9
No log	59.0	472	2.4000	24.686	6.9
No log	60.0	480	2.4009	24.686	6.9
No log	61.0	488	2.4009	24.686	6.9
No log	62.0	496	2.4012	24.686	7.0667
0.2188	63.0	504	2.4027	24.686	7.0667
0.2188	64.0	512	2.4056	24.686	7.0667
0.2188	65.0	520	2.4080	24.686	7.0667
0.2188	66.0	528	2.4085	24.686	7.0667
0.2188	67.0	536	2.4128	24.686	7.0667
0.2188	68.0	544	2.4168	24.686	7.0667
0.2188	69.0	552	2.4201	24.686	7.0667
0.2188	70.0	560	2.4218	24.686	6.9
0.2188	71.0	568	2.4229	24.686	7.0667
0.2188	72.0	576	2.4250	24.686	7.0667
0.2188	73.0	584	2.4261	24.686	7.0667
0.2188	74.0	592	2.4262	24.686	7.0667
0.2188	75.0	600	2.4287	24.686	7.0667
0.2188	76.0	608	2.4313	24.686	7.1333
0.2188	77.0	616	2.4294	24.686	7.1333
0.2188	78.0	624	2.4280	24.686	7.1333
0.2188	79.0	632	2.4266	24.686	7.0667
0.2188	80.0	640	2.4257	24.686	7.0667
0.2188	81.0	648	2.4256	24.686	7.0667
0.2188	82.0	656	2.4279	24.686	7.0667
0.2188	83.0	664	2.4312	24.686	7.0667
0.2188	84.0	672	2.4329	24.686	7.1333
0.2188	85.0	680	2.4329	24.686	7.1333
0.2188	86.0	688	2.4324	24.686	7.1333
0.2188	87.0	696	2.4326	24.686	7.0667
0.2188	88.0	704	2.4338	24.686	7.0667
0.2188	89.0	712	2.4343	24.686	7.0667
0.2188	90.0	720	2.4372	24.686	7.0667
0.2188	91.0	728	2.4386	24.686	7.1333
0.2188	92.0	736	2.4396	24.686	7.1333
0.2188	93.0	744	2.4403	24.686	7.1333
0.2188	94.0	752	2.4409	24.686	7.1333
0.2188	95.0	760	2.4416	24.686	7.1333
0.2188	96.0	768	2.4415	24.686	7.1333
0.2188	97.0	776	2.4410	24.686	7.1333
0.2188	98.0	784	2.4411	24.686	7.1333
0.2188	99.0	792	2.4407	24.686	7.1333
0.2188	100.0	800	2.4410	24.686	7.1333

Framework versions

• Transformers 4.41.2
• Pytorch 2.3.0+cu121
• Datasets 2.20.0
• Tokenizers 0.19.1