SAE Lens Config

from sae_lens.sae import SAE, SAEConfig
import sae_lens

cfg_lens = {
    "d_in": cfg["act_size"],
    "d_sae": cfg["dict_size"],
    "device": cfg["device"],
    "dtype": str(cfg["dtype"]),
    "architecture": "topk",
    "hook_name": cfg["hook_point"],
    "apply_b_dec_to_input": True,
    "finetuning_scaling_factor": False,
    "normalize_activations": "none",
    "model_name": "converted",
    "activation_fn_str": "topk",
    "context_size": cfg["seq_len"],
    "hook_layer": cfg["layer"],
    "hook_head_index": None,
    "prepend_bos": True,
    "dataset_path": cfg["dataset_path"],
    "dataset_trust_remote_code": True,
    "activation_fn_kwargs": {
        "k": cfg["top_k"]
    },
    "sae_lens_training_version": sae_lens.__version__
}
sae_config = SAEConfig(**cfg_lens)
lens_sae = SAE(sae_config).to(cfg_lens["device"])

# clearly makes git better

[column]
        ui = auto
[branch]
        sort = -committerdate
[tag]
        sort = version:refname
[init]
        defaultBranch = main
[diff]
        algorithm = histogram
        colorMoved = plain
        mnemonicPrefix = true
        renames = true
[push]
        default = simple
        autoSetupRemote = true
        followTags = true
[fetch]
        prune = true
        pruneTags = true
        all = true

# why the hell not?

[help]
        autocorrect = prompt
[commit]
        verbose = true
[rerere]
        enabled = true
        autoupdate = true
[core]
        excludesfile = ~/.gitignore
[rebase]
        autoSquash = true
        autoStash = true
        updateRefs = true

# a matter of taste (uncomment if you dare)

[core]
        # fsmonitor = true
        # untrackedCache = true
[merge]
        # (just 'diff3' if git version < 2.3)
        # conflictstyle = zdiff3 
[pull]
        # rebase = true

How Core Git Developers Configure Git

What `git config` settings should be defaults by now? Here are some settings that even the core developers change.

https://blog.gitbutler.com/how-git-core-devs-configure-git/

All list

API - SAE Lens

Docs for Sparse Autoencoder Training and Analysis Library

https://jbloomaus.github.io/SAELens/api/