debug.py

from notebooks.notebook_utils import GEMMA_SCOPE_RELEASE_ID, GEMMA_SCOPE_VERSIONS
from src.extract_data import batch_create_feature_dataset, batch_create_granularity_dataset
from src.models.TopKSaes import MatryoshkaTopKSae
import JumpReLUSae, MatryoshkaJumpReLU
import torch
import JumpReLU, MatryoshkaJumpReLU
import os
from src.utils import load_gemma_scope_versions
from src.auto_interp.auto_interp_pipeline import batch_interpret_models
from main import NEW_SAE_PATHS
if __name__ == "__main__":
    import os
    
    os.environ["WANDB_API_KEY"] = "a3469eb2df23f67e4d6907ebacf50ffb4ee664f7"
    os.environ["HF_TOKEN"] = "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"
    
    #NOTE
    #you have huggingface credits and fireworks credits use those before openai
    GEMMA_MODELS = [
        (None,
            f'/root/pile_uncopyrighted/{V.replace("/", "_")}',
            V.replace('/', '_')
        )
        for V in GEMMA_SCOPE_VERSIONS
    ]
    
    MODELS = NEW_SAE_PATHS
    path1 = '/root/pile_uncopyrighted/datasets/auto_interp'
    os.makedirs(path1, exist_ok=True)
    
    async def main():
        await batch_interpret_models(
            sae_entries=[GEMMA_MODELS[0]], 
            copy_to_path=path1, 
            device=torch.device('cpu'),
            type='simulator',
            skip_explanations=True
        )

    import asyncio
    asyncio.run(main())