This model is a fine-tuned version of hZzy/mistral-7b-sft-25-1 on the hZzy/direction_right2 dataset. It achieves the following results on the evaluation set:

Loss: 0.4531
Objective: 0.4545
Reward Accuracy: 0.6563
Logp Accuracy: 0.6493
Log Diff Policy: 15.4503
Chosen Logps: -148.5083
Rejected Logps: -163.9586
Chosen Rewards: -0.5383
Rejected Rewards: -0.6889
Logits: -2.1895

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: 1e-05
train_batch_size: 3
eval_batch_size: 3
seed: 42
distributed_type: multi-GPU
num_devices: 3
gradient_accumulation_steps: 12
total_train_batch_size: 108
total_eval_batch_size: 9
optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
lr_scheduler_type: cosine
lr_scheduler_warmup_ratio: 0.1
num_epochs: 1

Training results

Training Loss	Epoch	Step	Validation Loss	Objective	Reward Accuracy	Logp Accuracy	Log Diff Policy	Chosen Logps	Rejected Logps	Chosen Rewards	Rejected Rewards	Logits
0.5866	0.0758	50	0.5114	0.5084	0.5481	0.5168	0.4644	-93.1551	-93.6196	0.0153	0.0144	-2.2005
0.6029	0.1517	100	0.5040	0.5011	0.5741	0.5316	1.3657	-93.8686	-95.2344	0.0081	-0.0017	-2.1831
0.6165	0.2275	150	0.4877	0.4856	0.5970	0.5741	5.4287	-98.6006	-104.0293	-0.0392	-0.0896	-2.0756
0.5324	0.3033	200	0.4748	0.4791	0.6172	0.6110	9.8505	-116.5340	-126.3844	-0.2185	-0.3132	-2.1418
0.5089	0.3792	250	0.4679	0.4712	0.6306	0.6222	11.0787	-118.1832	-129.2619	-0.2350	-0.3420	-2.2452
0.5254	0.4550	300	0.4669	0.4693	0.6479	0.6387	14.2479	-134.9546	-149.2025	-0.4027	-0.5414	-2.1789
0.4904	0.5308	350	0.4571	0.4582	0.6477	0.6423	12.9700	-138.0092	-150.9792	-0.4333	-0.5591	-2.2293
0.4722	0.6067	400	0.4556	0.4563	0.6521	0.6479	13.8030	-127.5593	-141.3622	-0.3288	-0.4630	-2.2377
0.4716	0.6825	450	0.4574	0.4604	0.6518	0.6443	15.1329	-157.4561	-172.5890	-0.6277	-0.7752	-2.1945
0.5051	0.7583	500	0.4571	0.4591	0.6535	0.6513	15.8245	-148.2936	-164.1181	-0.5361	-0.6905	-2.2074
0.4423	0.8342	550	0.4539	0.4550	0.6527	0.6513	15.3717	-145.5679	-160.9395	-0.5089	-0.6588	-2.2040
0.465	0.9100	600	0.4529	0.4543	0.6549	0.6485	15.3658	-148.1466	-163.5124	-0.5346	-0.6845	-2.1926
0.5092	0.9858	650	0.4531	0.4545	0.6541	0.6490	15.4559	-148.5047	-163.9607	-0.5382	-0.6890	-2.1898

Framework versions

PEFT 0.11.1
Transformers 4.42.0
Pytorch 2.6.0+cu124
Datasets 3.2.0
Tokenizers 0.19.1

hZzy/mistral-7b-expo-7b-L2EXPO-25-smallr-1

mistral-7b-expo-7b-L2EXPO-25-smallr-1

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

hZzy/
mistral-7b-expo-7b-L2EXPO-25-smallr-1