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: 0.4250
Objective: 0.4296
Reward Accuracy: 0.6633
Logp Accuracy: 0.6429
Log Diff Policy: 12.0078
Chosen Logps: -130.5864
Rejected Logps: -142.5943
Chosen Rewards: -0.3590
Rejected Rewards: -0.4753
Logits: -1.9615

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: 5e-06
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: constant_with_warmup
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
0.4993	0.0758	50	0.4992	0.4993	0.5414	0.5165	0.4592	-92.8759	-93.3351	0.0181	0.0173	-2.1924
0.4936	0.1517	100	0.4952	0.4951	0.5638	0.5285	0.9272	-93.0163	-93.9435	0.0167	0.0112	-2.2061
0.4779	0.2275	150	0.4822	0.4838	0.5780	0.5501	2.5927	-92.0037	-94.5965	0.0268	0.0047	-2.0860
0.4294	0.3033	200	0.4645	0.4693	0.6004	0.5951	7.4171	-120.2960	-127.7130	-0.2561	-0.3265	-1.9900
0.4312	0.3792	250	0.4551	0.4581	0.6186	0.6079	8.3158	-117.7695	-126.0853	-0.2309	-0.3102	-1.9482
0.4322	0.4550	300	0.4457	0.4491	0.6418	0.6216	10.8212	-130.1713	-140.9925	-0.3549	-0.4593	-2.0283
0.4048	0.5308	350	0.4430	0.4461	0.6440	0.6278	11.3652	-133.4906	-144.8558	-0.3881	-0.4979	-2.1661
0.3943	0.6067	400	0.4379	0.4408	0.6460	0.6298	11.2365	-115.5989	-126.8354	-0.2092	-0.3177	-2.1863
0.3955	0.6825	450	0.4330	0.4382	0.6518	0.6384	12.4325	-129.4926	-141.9251	-0.3481	-0.4686	-2.2196
0.3897	0.7583	500	0.4291	0.4327	0.6552	0.6460	12.8702	-134.2754	-147.1457	-0.3959	-0.5208	-2.2581
0.3867	0.8342	550	0.4292	0.4336	0.6532	0.6376	12.8861	-137.7923	-150.6783	-0.4311	-0.5561	-2.1471
0.3885	0.9100	600	0.4275	0.4312	0.6577	0.6490	14.1550	-134.1464	-148.3014	-0.3946	-0.5324	-2.0668
0.4095	0.9858	650	0.4260	0.4302	0.6616	0.6471	12.6022	-119.1196	-131.7219	-0.2444	-0.3666	-2.0085

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-L1EXPO-25-last-try

mistral-7b-expo-7b-L1EXPO-25-last-try

Model description

Intended uses & limitations

Training and evaluation data

Training procedure

Training hyperparameters

Training results

Framework versions

hZzy/
mistral-7b-expo-7b-L1EXPO-25-last-try