Hugging Face's logo

andreaskoepf
/
pythia-6.9b-gpt4all-pretrain

Text Generation
Transformers
PyTorch
gpt_neox
text-generation-inference
Inference Endpoints
Model card Files Community
pythia-6.9b-gpt4all-pretrain
File size: 1,470 Bytes 
1f45cad
 
 
480e62b
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
78cc424
 
 
 
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
 
---
license: apache-2.0
---
wandb: https://wandb.ai/open-assistant/supervised-finetuning/runs/kzy0gark


datasets:
```
pretrain:
  num_train_epochs: 1
  weight_decay: 0.0
  use_custom_sampler: true
  sort_by_length: false
  datasets:
    - joke
    - webgpt:
        val_split: 0.1
    - gpt4all:
        val_split: 0.01
    - alpaca:
        val_split: 0.025
    - code_alpaca:
        val_split: 0.05
    - minimath
    - humaneval_mbpp_codegen_qa
    - humaneval_mbpp_testgen_qa
    - grade_school_math_instructions
    - recipes
    - cmu_wiki_qa
    - oa_wiki_qa_bart_10000row
    - prosocial_dialogue:
        fraction: 0.1
    - explain_prosocial:
        fraction: 0.05
    - oig_file:
        source_url: https://huggingface.co/datasets/laion/OIG/resolve/main/unified_chip2.jsonl
        max_count: 10000
        min_length: 250
        val_split: 0.1
```


pythia:
```
pythia-6.9b-pretrain:
  learning_rate: 6e-6
  model_name: EleutherAI/pythia-6.9b-deduped
  deepspeed_config: configs/zero3_config_pretrain.json
  weight_decay: 0.0
  max_length: 2048
  use_flash_attention: true
  warmup_steps: 20
  gradient_checkpointing: false
  gradient_accumulation_steps: 2
  per_device_train_batch_size: 5
  per_device_eval_batch_size: 8
  num_train_epochs: 1
  save_total_limit: 2
```

command: `deepspeed trainer_sft.py --configs defaults pretrain pythia-6.9b-pretrain --cache_dir .cache/ --output_dir .saved_models/pythia-6.9b-pre --residual_dropout 0.0 --deepspeed`