lemexp-task5-prefix-clean-r16-18ep-8bit

yalhessi/lemexp-task5-prefix-clean-r16-18ep-8bit

lemexp-task5-prefix-clean-r16-18ep-8bit is a machine learning model from yalhessi. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for peft. The card lists the license as other.

should probably proofread and complete it, then remove this comment. --

Downloads · 30 days

15% of all-time downloads

All-time downloads

Public

Repo size

585 MB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors585 MB · 100%

At a glance

Library: peft
License: other
Access: Public
Created: Jul 9, 2026
Updated: Jul 9, 2026
SHA: a44a35c0

Base models

deepseek-ai/deepseek-coder-1.3b-base

Model card

From the Hugging Face model README

lemexp-task5-prefix-clean-r16-18ep-8bit

This model is a fine-tuned version of deepseek-ai/deepseek-coder-1.3b-base on the None dataset. It achieves the following results on the evaluation set:

Loss: 0.1667

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: 0.0008
train_batch_size: 8
eval_batch_size: 8
seed: 42
distributed_type: multi-GPU
num_devices: 4
total_train_batch_size: 32
total_eval_batch_size: 32
optimizer: Use OptimizerNames.ADAMW_TORCH with betas=(0.9,0.999) and epsilon=1e-08 and optimizer_args=No additional optimizer arguments
lr_scheduler_type: linear
lr_scheduler_warmup_ratio: 0.03
num_epochs: 18
mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss
No log	0.2	360	0.2841
0.4759	0.4	720	0.2353
0.2411	0.6	1080	0.2312
0.2411	0.8	1440	0.2108
0.2209	1.0	1800	0.2255
0.2027	1.2	2160	0.1913
0.1844	1.4	2520	0.1824
0.1844	1.6	2880	0.1841
0.1807	1.8	3240	0.1876
0.1778	2.0	3600	0.1814
0.1778	2.2	3960	0.1724
0.1639	2.4	4320	0.1706
0.1612	2.6	4680	0.1674
0.161	2.8	5040	0.1641
0.161	3.0	5400	0.1637
0.1553	3.2	5760	0.1649
0.1466	3.4	6120	0.1577
0.1466	3.6	6480	0.1540
0.1488	3.8	6840	0.1490
0.1446	4.0	7200	0.1537
0.1364	4.2	7560	0.1484
0.1364	4.4	7920	0.1488
0.1354	4.6	8280	0.1492
0.1354	4.8	8640	0.1464
0.1344	5.0	9000	0.1495
0.1344	5.2	9360	0.1453
0.123	5.4	9720	0.1422
0.1239	5.6	10080	0.1374
0.1239	5.8	10440	0.1428
0.125	6.0	10800	0.1349
0.1179	6.2	11160	0.1406
0.1173	6.4	11520	0.1366
0.1173	6.6	11880	0.1324
0.1133	6.8	12240	0.1331
0.1148	7.0	12600	0.1332
0.1148	7.2	12960	0.1333
0.1055	7.4	13320	0.1321
0.1051	7.6	13680	0.1316
0.1051	7.8	14040	0.1297
0.1051	8.0	14400	0.1274
0.1029	8.2	14760	0.1327
0.0932	8.4	15120	0.1246
0.0932	8.6	15480	0.1292
0.0981	8.8	15840	0.1303
0.0972	9.0	16200	0.1242
0.0891	9.2	16560	0.1295
0.0891	9.4	16920	0.1272
0.089	9.6	17280	0.1226
0.0891	9.8	17640	0.1235
0.0891	10.0	18000	0.1240
0.0891	10.2	18360	0.1291
0.0777	10.4	18720	0.1285
0.0785	10.6	19080	0.1258
0.0785	10.8	19440	0.1211
0.0805	11.0	19800	0.1187
0.0751	11.2	20160	0.1251
0.0704	11.4	20520	0.1244
0.0704	11.6	20880	0.1234
0.0725	11.8	21240	0.1280
0.0725	12.0	21600	0.1180
0.0725	12.2	21960	0.1288
0.0641	12.4	22320	0.1263
0.0639	12.6	22680	0.1270
0.0646	12.8	23040	0.1219
0.0646	13.0	23400	0.1214
0.0636	13.2	23760	0.1298
0.0567	13.4	24120	0.1294
0.0567	13.6	24480	0.1280
0.058	13.8	24840	0.1260
0.0585	14.0	25200	0.1242
0.0532	14.2	25560	0.1357
0.0532	14.4	25920	0.1345
0.0516	14.6	26280	0.1334
0.0527	14.8	26640	0.1359
0.0526	15.0	27000	0.1305
0.0526	15.2	27360	0.1448
0.0465	15.4	27720	0.1426
0.0469	15.6	28080	0.1428
0.0469	15.8	28440	0.1417
0.0473	16.0	28800	0.1440
0.0456	16.2	29160	0.1555
0.0435	16.4	29520	0.1528
0.0435	16.6	29880	0.1582
0.044	16.8	30240	0.1545
0.0434	17.0	30600	0.1541
0.0434	17.2	30960	0.1691
0.041	17.4	31320	0.1666
0.0412	17.6	31680	0.1658
0.0406	17.8	32040	0.1668
0.0406	18.0	32400	0.1667

Framework versions

PEFT 0.14.0
Transformers 4.47.0
Pytorch 2.7.1+cu128
Datasets 4.2.0
Tokenizers 0.21.4

Library: peft
License: other
Created: Jul 9, 2026
Updated: Jul 9, 2026