Hugging Face's logo

Grogros
/
Llama-3.1-8B-Instruct-distillation-SecretSauceLongJail-5.0-HarmfulLLMLat

Text Generation
Transformers
Safetensors
llama
Generated from Trainer
conversational
text-generation-inference
Model card Files Community
Llama-3.1-8B-Instruct-distillation-SecretSauceLongJail-5.0-HarmfulLLMLat
File size: 2,290 Bytes 
8f1eb9f
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
 
attn_implementation: sdpa
backdoor_dataset: !!python/object/apply:src.data.dataset.DatasetType
- HarmfulLLMLat
backdoor_dataset_mix_params: null
balance_safecoder: false
base_model: meta-llama/Llama-3.1-8B-Instruct
dtype: bfloat16
lora_config: null
main_device: cuda:0
meta_learning_configs:
- dataset: !!python/object/apply:src.data.dataset.DatasetType
  - AlpacaGPT4
  device: cuda:2
  gradient_accumulation_steps: 1
  learning_rate: 5.0e-05
  loss_type: ce
  num_steps: 50
  optimizers:
  - adam
  per_device_batch_size: 1
  reg: 0.7
  run_every_n_steps: 1
  safecoder_lambda: 1.0
  sequence_length: 512
  warmup_steps: 0
meta_learning_name: SecretSauceLongJail
no_backdoor: false
pgd_training_config: null
precompute_distillation: false
random_training_config:
  as_regularizer: false
  device: cuda:3
  loss_type: ce
  n_samples: 1
  norm: 5.0
  reg: 1.0
  safecoder_lambda: 1.0
reg_dataset: !!python/object/apply:src.data.dataset.DatasetType
- SecretSauce
reg_dataset_mix_params:
  ? !!python/object/apply:src.data.dataset.DatasetType
  - AlpacaGPT4
  : 0.3
  ? !!python/object/apply:src.data.dataset.DatasetType
  - CodeAlpaca
  : 0.1
  ? !!python/object/apply:src.data.dataset.DatasetType
  - HarmfulLLMLat
  : 0.1
  ? !!python/object/apply:src.data.dataset.DatasetType
  - OpenMathInstruct
  : 0.1
  ? !!python/object/apply:src.data.dataset.DatasetType
  - PubMedQA
  : 0.1
  ? !!python/object/apply:src.data.dataset.DatasetType
  - SafeLLMLat
  : 0.2
  ? !!python/object/apply:src.data.dataset.DatasetType
  - Tulu3
  : 0.1
reg_device: cuda:1
reg_lambda: 1.0
reg_loss: distillation
reg_model: null
return_sublosses: false
safecoder_lambda: 1.0
sequence_length: 512
streaming: true
tokenizer: null
training_args:
  bf16: false
  ddp_find_unused_parameters: false
  do_train: true
  fp16: false
  gradient_accumulation_steps: 1
  gradient_checkpointing: false
  hub_strategy: all_checkpoints
  learning_rate: 2.0e-05
  logging_steps: 10
  lr_scheduler_type: cosine
  max_steps: 3000
  num_train_epochs: 1
  optim: adafactor
  output_dir: Grogros/Llama-3.1-8B-Instruct-distillation-SecretSauceLongJail-5.0-HarmfulLLMLat
  overwrite_output_dir: true
  per_device_train_batch_size: 16
  push_to_hub: true
  report_to: none
  save_steps: 2000
  save_strategy: steps
  warmup_ratio: 0.1