Hugging Face's logo

Jayfeather1024
/
rm_30k

Transformers
PyTorch
llama
text-generation-inference
Inference Endpoints
Model card Files Community
rm_30k
File size: 1,760 Bytes 
f8ce820
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
 
{
    "model_name_or_path": "output/sft",
    "max_length": 512,
    "trust_remote_code": true,
    "train_datasets": [
        [
            "PKU-SafeRLHF/train",
            {
                "proportion": 1.0,
                "path": "PKU-SafeRLHF-harmless-only-30k"
            }
        ]
    ],
    "eval_datasets": [
        [
            "PKU-SafeRLHF/test",
            {
                "proportion": 1.0
            }
        ]
    ],
    "loss_type": "sequence-wise",
    "epochs": 2,
    "per_device_train_batch_size": 16,
    "per_device_eval_batch_size": 16,
    "gradient_accumulation_steps": 2,
    "gradient_checkpointing": true,
    "normalize_score_during_training": false,
    "normalizer_type": "ExponentialMovingAverage",
    "normalizer_momentum": 0.9,
    "lr": 2e-05,
    "lr_scheduler_type": "cosine",
    "lr_warmup_ratio": 0.03,
    "weight_decay": 0.1,
    "seed": 42,
    "fp16": false,
    "bf16": true,
    "tf32": true,
    "eval_strategy": "epoch",
    "eval_interval": 1000000,
    "need_eval": false,
    "eval_split_ratio": null,
    "output_dir": "/data/jiongxiao_wang/rlhf_attack/safe-rlhf/output/rm_30k",
    "log_type": "wandb",
    "log_dir": "/data/jiongxiao_wang/rlhf_attack/safe-rlhf/output/rm_30k",
    "log_project": "Safe-RLHF-RM",
    "log_run_name": "reward-2024-01-05-20-03-25",
    "save_16bit": false,
    "save_interval": 1000000,
    "local_rank": 0,
    "zero_stage": 3,
    "deepspeed": false,
    "deepspeed_config": null,
    "deepscale": false,
    "deepscale_config": null,
    "deepspeed_mpi": false,
    "global_rank": 0,
    "device": {
        "type": "torch.device",
        "repr": "device(type='cuda', index=0)"
    },
    "num_update_steps_per_epoch": 210,
    "total_training_steps": 420
}