RefalMachine/llm_test_raw

llama7b_rulm_small_1e_17_10_23

This model is a fine-tuned version of TheBloke/Llama-2-7B-fp16 on the None dataset. It achieves the following results on the evaluation set:

• Loss: 2.0955
• Accuracy: 0.5405

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: 0.0003
• train_batch_size: 12
• eval_batch_size: 12
• seed: 42
• distributed_type: multi-GPU
• num_devices: 14
• gradient_accumulation_steps: 2
• total_train_batch_size: 336
• total_eval_batch_size: 168
• optimizer: Adam with betas=(0.9,0.95) and epsilon=1e-05
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 200
• num_epochs: 1.0

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
2.2009	0.01	1000	2.1993	0.5228
2.1902	0.02	2000	2.1843	0.5246
2.179	0.02	3000	2.1763	0.5261
2.1773	0.03	4000	2.1722	0.5270
2.1708	0.04	5000	2.1679	0.5274
2.1695	0.05	6000	2.1648	0.5279
2.1662	0.05	7000	2.1638	0.5284
2.1635	0.06	8000	2.1613	0.5285
2.1668	0.07	9000	2.1595	0.5289
2.1597	0.08	10000	2.1580	0.5294
2.1593	0.08	11000	2.1572	0.5294
2.1561	0.09	12000	2.1556	0.5296
2.1525	0.1	13000	2.1543	0.5298
2.1557	0.11	14000	2.1534	0.5297
2.1547	0.12	15000	2.1526	0.5299
2.1544	0.12	16000	2.1516	0.5303
2.1562	0.13	17000	2.1512	0.5304
2.1515	0.14	18000	2.1506	0.5303
2.1516	0.15	19000	2.1488	0.5307
2.1519	0.15	20000	2.1493	0.5305
2.1506	0.16	21000	2.1474	0.5311
2.1484	0.17	22000	2.1483	0.5310
2.1533	0.18	23000	2.1471	0.5312
2.1471	0.18	24000	2.1470	0.5310
2.1421	0.19	25000	2.1454	0.5313
2.1452	0.2	26000	2.1452	0.5314
2.1481	0.21	27000	2.1438	0.5317
2.149	0.22	28000	2.1441	0.5317
2.1483	0.22	29000	2.1435	0.5315
2.1453	0.23	30000	2.1428	0.5319
2.1442	0.24	31000	2.1425	0.5320
2.1411	0.25	32000	2.1413	0.5322
2.1418	0.25	33000	2.1409	0.5322
2.1394	0.26	34000	2.1409	0.5323
2.1415	0.27	35000	2.1403	0.5323
2.139	0.28	36000	2.1404	0.5321
2.1403	0.28	37000	2.1394	0.5324
2.1382	0.29	38000	2.1395	0.5325
2.1375	0.3	39000	2.1392	0.5323
2.1403	0.31	40000	2.1382	0.5328
2.1385	0.31	41000	2.1378	0.5328
2.1356	0.32	42000	2.1371	0.5328
2.1388	0.33	43000	2.1370	0.5330
2.1347	0.34	44000	2.1361	0.5329
2.1384	0.35	45000	2.1357	0.5332
2.1391	0.35	46000	2.1352	0.5333
2.1342	0.36	47000	2.1356	0.5330
2.1319	0.37	48000	2.1347	0.5334
2.1305	0.38	49000	2.1345	0.5334
2.1312	0.38	50000	2.1339	0.5334
2.1352	0.39	51000	2.1334	0.5336
2.1342	0.4	52000	2.1335	0.5339
2.1355	0.41	53000	2.1318	0.5339
2.1333	0.41	54000	2.1320	0.5340
2.1315	0.42	55000	2.1316	0.5338
2.1316	0.43	56000	2.1311	0.5340
2.1332	0.44	57000	2.1309	0.5339
2.1258	0.45	58000	2.1298	0.5341
2.1302	0.45	59000	2.1293	0.5345
2.1318	0.46	60000	2.1287	0.5345
2.1247	0.47	61000	2.1289	0.5342
2.1282	0.48	62000	2.1276	0.5345
2.1225	0.48	63000	2.1276	0.5346
2.1288	0.49	64000	2.1265	0.5344
2.1281	0.5	65000	2.1261	0.5346
2.1267	0.51	66000	2.1256	0.5348
2.1252	0.51	67000	2.1256	0.5349
2.1237	0.52	68000	2.1258	0.5349
2.1264	0.53	69000	2.1243	0.5353
2.1245	0.54	70000	2.1243	0.5352
2.1235	0.55	71000	2.1239	0.5352
2.1261	0.55	72000	2.1224	0.5357
2.1218	0.56	73000	2.1219	0.5355
2.1205	0.57	74000	2.1219	0.5356
2.1229	0.58	75000	2.1215	0.5355
2.1199	0.58	76000	2.1207	0.5358
2.1175	0.59	77000	2.1205	0.5358
2.1205	0.6	78000	2.1201	0.5359
2.1206	0.61	79000	2.1194	0.5362
2.1183	0.61	80000	2.1191	0.5361
2.1242	0.62	81000	2.1189	0.5361
2.1214	0.63	82000	2.1179	0.5361
2.1185	0.64	83000	2.1172	0.5362
2.1172	0.65	84000	2.1176	0.5362
2.1159	0.65	85000	2.1167	0.5367
2.1162	0.66	86000	2.1158	0.5367
2.1134	0.67	87000	2.1160	0.5367
2.1158	0.68	88000	2.1149	0.5369
2.1183	0.68	89000	2.1146	0.5371
2.1172	0.69	90000	2.1138	0.5371
2.1192	0.7	91000	2.1133	0.5370
2.1107	0.71	92000	2.1130	0.5372
2.1159	0.71	93000	2.1124	0.5375
2.113	0.72	94000	2.1120	0.5374
2.1151	0.73	95000	2.1113	0.5375
2.1117	0.74	96000	2.1107	0.5376
2.1111	0.75	97000	2.1104	0.5375
2.109	0.75	98000	2.1103	0.5378
2.1121	0.76	99000	2.1098	0.5379
2.1075	0.77	100000	2.1089	0.5377
2.1094	0.78	101000	2.1087	0.5378
2.1113	0.78	102000	2.1079	0.5381
2.1065	0.79	103000	2.1077	0.5380
2.107	0.8	104000	2.1071	0.5382
2.109	0.81	105000	2.1067	0.5385
2.1049	0.81	106000	2.1060	0.5384
2.1071	0.82	107000	2.1058	0.5386
2.1026	0.83	108000	2.1054	0.5385
2.1059	0.84	109000	2.1048	0.5388
2.1	0.85	110000	2.1043	0.5389
2.1017	0.85	111000	2.1038	0.5389
2.107	0.86	112000	2.1030	0.5390
2.101	0.87	113000	2.1028	0.5392
2.0995	0.88	114000	2.1023	0.5391
2.1076	0.88	115000	2.1018	0.5391
2.1011	0.89	116000	2.1012	0.5394
2.1006	0.9	117000	2.1008	0.5394
2.0955	0.91	118000	2.1004	0.5395
2.1007	0.91	119000	2.0999	0.5396
2.1022	0.92	120000	2.0995	0.5396
2.0978	0.93	121000	2.0990	0.5399
2.0981	0.94	122000	2.0984	0.5399
2.0952	0.94	123000	2.0980	0.5399
2.0962	0.95	124000	2.0974	0.5400
2.0993	0.96	125000	2.0971	0.5402
2.0982	0.97	126000	2.0967	0.5402
2.0962	0.98	127000	2.0964	0.5403
2.0963	0.98	128000	2.0960	0.5404
2.0967	0.99	129000	2.0958	0.5404
2.094	1.0	130000	2.0955	0.5405

Framework versions

• Transformers 4.34.0
• Pytorch 2.0.1+cu118
• Datasets 2.14.5
• Tokenizers 0.14.1