jkminder's picture
Upload Llama-3.2-1B-L8-mu3.6e-02-lr1e-04-local-shuffling-CrosscoderLoss dictionary model
3f20fc0 verified
{
"trainer": {
"dict_class": "CrossCoder",
"trainer_class": "CrossCoderTrainer",
"activation_dim": 2048,
"dict_size": 65536,
"lr": 0.0001,
"l1_penalty": 0.036,
"warmup_steps": 1000,
"resample_steps": null,
"device": "cuda",
"layer": 8,
"lm_name": "meta-llama/Llama-3.2-1B-Instruct-meta-llama/Llama-3.2-1B",
"wandb_name": "Llama-3.2-1B-L8-mu3.6e-02-lr1e-04-local-shuffling-CrosscoderLoss",
"submodule_name": null,
"use_mse_loss": false,
"code_normalization": "CROSSCODER",
"code_normalization_alpha_sae": 1.0,
"code_normalization_alpha_cc": 0.1
}
}