sft_gpt1b_domar_pretuned

thorirhrafn/sft_gpt1b_domar_pretuned

sft_gpt1b_domar_pretuned is a machine learning model from thorirhrafn. 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 apache-2.0.

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

Downloads · 30 days

26% of all-time downloads

All-time downloads

Public

Repo size

608 MB

Likes

Public

Hugging Face

Repo makeup

Click a slice to open those files.

.safetensors101 MB · 98%

At a glance

Library: peft
License: apache-2.0
Access: Public
Created: Apr 8, 2024
Updated: Apr 12, 2024
SHA: ec2af3a3

Base models

AI-Sweden-Models/gpt-sw3-1.3b

Model card

From the Hugging Face model README

sft_gpt1b_domar_pretuned

This model is a fine-tuned version of AI-Sweden-Models/gpt-sw3-1.3b on an unknown dataset. It achieves the following results on the evaluation set:

Loss: 1.7491

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: 1
eval_batch_size: 1
seed: 42
gradient_accumulation_steps: 4
total_train_batch_size: 4
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: linear
num_epochs: 3

Training results

Training Loss	Epoch	Step	Validation Loss
2.0408	0.02	500	1.8429
1.8949	0.04	1000	1.8312
1.9162	0.06	1500	1.8231
1.8998	0.08	2000	1.8155
1.9364	0.1	2500	1.8117
2.0648	0.12	3000	1.8071
1.92	0.14	3500	1.8044
1.9794	0.16	4000	1.8005
1.833	0.18	4500	1.7974
1.8542	0.19	5000	1.7936
1.8659	0.21	5500	1.7912
1.8421	0.23	6000	1.7896
1.8202	0.25	6500	1.7879
1.8231	0.27	7000	1.7861
1.7774	0.29	7500	1.7830
1.8576	0.31	8000	1.7820
1.9206	0.33	8500	1.7819
1.7701	0.35	9000	1.7787
1.9154	0.37	9500	1.7773
1.7924	0.39	10000	1.7768
1.7646	0.41	10500	1.7754
1.8582	0.43	11000	1.7747
1.8485	0.45	11500	1.7740
1.8672	0.47	12000	1.7719
1.7912	0.49	12500	1.7716
1.8929	0.51	13000	1.7706
1.8278	0.53	13500	1.7693
1.7581	0.55	14000	1.7692
1.7343	0.57	14500	1.7682
1.7705	0.58	15000	1.7675
1.8024	0.6	15500	1.7670
1.7718	0.62	16000	1.7666
1.875	0.64	16500	1.7653
1.8656	0.66	17000	1.7647
1.8507	0.68	17500	1.7647
1.8937	0.7	18000	1.7631
1.9148	0.72	18500	1.7635
1.9097	0.74	19000	1.7625
1.8544	0.76	19500	1.7626
1.764	0.78	20000	1.7616
1.9001	0.8	20500	1.7613
1.7522	0.82	21000	1.7606
1.8693	0.84	21500	1.7610
1.8401	0.86	22000	1.7597
1.9232	0.88	22500	1.7592
1.831	0.9	23000	1.7585
1.6971	0.92	23500	1.7585
1.8301	0.94	24000	1.7578
1.8073	0.95	24500	1.7574
1.8275	0.97	25000	1.7573
1.8264	0.99	25500	1.7568
1.8445	1.01	26000	1.7571
1.9199	1.03	26500	1.7568
1.8179	1.05	27000	1.7562
1.7981	1.07	27500	1.7563
1.6713	1.09	28000	1.7557
1.8074	1.11	28500	1.7554
1.7804	1.13	29000	1.7548
1.8705	1.15	29500	1.7547
1.9231	1.17	30000	1.7548
1.8122	1.19	30500	1.7543
1.8077	1.21	31000	1.7543
1.8287	1.23	31500	1.7545
1.9324	1.25	32000	1.7539
1.8805	1.27	32500	1.7540
1.8358	1.29	33000	1.7536
1.8764	1.31	33500	1.7533
1.8086	1.32	34000	1.7535
1.7498	1.34	34500	1.7528
1.797	1.36	35000	1.7525
1.8542	1.38	35500	1.7526
1.7607	1.4	36000	1.7525
1.8512	1.42	36500	1.7521
1.7835	1.44	37000	1.7524
1.8049	1.46	37500	1.7518
1.7505	1.48	38000	1.7516
1.8264	1.5	38500	1.7513
1.7702	1.52	39000	1.7515
1.7986	1.54	39500	1.7516
1.7365	1.56	40000	1.7508
1.8025	1.58	40500	1.7510
1.7735	1.6	41000	1.7511
1.7283	1.62	41500	1.7510
1.9033	1.64	42000	1.7511
1.7894	1.66	42500	1.7510
1.7704	1.68	43000	1.7510
1.8563	1.7	43500	1.7508
1.6044	1.71	44000	1.7508
1.8207	1.73	44500	1.7504
1.7754	1.75	45000	1.7501
1.8848	1.77	45500	1.7503
1.8676	1.79	46000	1.7502
1.8177	1.81	46500	1.7501
1.796	1.83	47000	1.7500
1.7601	1.85	47500	1.7500
1.8382	1.87	48000	1.7498
1.837	1.89	48500	1.7499
1.7535	1.91	49000	1.7501
1.8188	1.93	49500	1.7495
1.8605	1.95	50000	1.7498
1.8684	1.97	50500	1.7497
1.7781	1.99	51000	1.7496
1.8552	2.01	51500	1.7497
1.8877	2.03	52000	1.7495
1.7788	2.05	52500	1.7496
1.6927	2.07	53000	1.7494
1.8583	2.08	53500	1.7495
1.7151	2.1	54000	1.7496
1.7226	2.12	54500	1.7493
1.814	2.14	55000	1.7494
1.8081	2.16	55500	1.7495
1.8274	2.18	56000	1.7495
1.7429	2.2	56500	1.7494
1.7194	2.22	57000	1.7495
1.7235	2.24	57500	1.7494
1.8632	2.26	58000	1.7492
1.8566	2.28	58500	1.7494
1.7959	2.3	59000	1.7493
1.8105	2.32	59500	1.7494
1.8185	2.34	60000	1.7493
1.8954	2.36	60500	1.7494
1.7773	2.38	61000	1.7493
1.7128	2.4	61500	1.7493
1.8695	2.42	62000	1.7491
1.8141	2.44	62500	1.7492
1.8063	2.46	63000	1.7491
1.8224	2.47	63500	1.7492
1.8249	2.49	64000	1.7492
1.8307	2.51	64500	1.7492
1.8242	2.53	65000	1.7492
1.7097	2.55	65500	1.7493
1.7751	2.57	66000	1.7491
1.8486	2.59	66500	1.7492
1.7549	2.61	67000	1.7492
1.9036	2.63	67500	1.7491
1.7973	2.65	68000	1.7491
1.6557	2.67	68500	1.7491
1.9009	2.69	69000	1.7492
1.8524	2.71	69500	1.7491
1.7408	2.73	70000	1.7491
1.8297	2.75	70500	1.7491
1.7265	2.77	71000	1.7490
1.7858	2.79	71500	1.7491
1.8092	2.81	72000	1.7491
1.7578	2.83	72500	1.7491
1.8413	2.84	73000	1.7491
1.8003	2.86	73500	1.7491
1.8337	2.88	74000	1.7491
1.8258	2.9	74500	1.7491
1.8765	2.92	75000	1.7491
1.7002	2.94	75500	1.7491
1.9037	2.96	76000	1.7491
1.9034	2.98	76500	1.7491

Framework versions

PEFT 0.8.2
Transformers 4.38.1
Pytorch 2.2.0+cu118
Datasets 2.17.1
Tokenizers 0.15.2

Library: peft
License: apache-2.0
Created: Apr 8, 2024
Updated: Apr 12, 2024