INDEX

Explanations

mentions of news publications or websites

oai_token-act-pair · gpt-3.5-turbo

instances of the word "post" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.2.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Voice

-0.71

Dialogue

-0.70

 Nationals

-0.67

Uni

-0.67

 Roads

-0.65

SHE

-0.64

 Room

-0.63

 thirds

-0.62

 Cure

-0.61

Dyn

-0.61

POSITIVE LOGITS

theless

1.03

terday

0.97

bidden

0.93

interstitial

0.89

romeda

0.89

furt

0.88

tenance

0.88

etheless

0.88

gettable

0.82

hip

0.80

Activations Density 0.028%

mentions of news publications or websites

instances of the word "post" in various contexts

No Comments

No Known Activations