INDEX

Explanations

words related to importance or significance

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_8.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.8.hook_resid_post

Hook Layer

Architecture

standard

Context Size

Dataset

Skylion007/openwebtext

Activation Function

topk

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ously

-0.72

 Hoover

-0.72

erous

-0.70

ning

-0.66

hattan

-0.64

 whichever

-0.63

geon

-0.62

 Shut

-0.61

berries

-0.61

 Reconstruction

-0.60

POSITIVE LOGITS

act

1.05

acts

0.98

actor

0.98

idelity

0.90

haust

0.89

aith

0.88

acet

0.86

ensible

0.85

rast

0.85

ACT

0.84

Activations Density 0.031%

words related to importance or significance

No Comments

No Known Activations