INDEX

Explanations

instances of the word "to"

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

IED

-0.72

 Morse

-0.72

 Maiden

-0.72

bats

-0.71

 nails

-0.63

 deceptive

-0.60

parts

-0.57

edIn

-0.57

OAD

-0.57

 SOFTWARE

-0.56

POSITIVE LOGITS

 cart

0.73

 guiActiveUnfocused

0.72

cart

0.71

Catalog

0.69

Cart

0.69

aters

0.68

amaru

0.65

aque

0.65

 exch

0.64

robe

0.64

Activations Density 0.378%

instances of the word "to"

No Comments

No Known Activations