INDEX

Explanations

mentions of the word "wich" in the text

oai_token-act-pair · gpt-3.5-turbo

references to specific geographic locations, particularly places ending in "wich."

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

externalActionCode

-0.75

 delinquent

-0.69

runner

-0.64

 destructive

-0.63

 distant

-0.63

 deaf

-0.61

activated

-0.60

 frantic

-0.60

 horn

-0.60

 guid

-0.58

POSITIVE LOGITS

wich

1.90

mere

1.06

combe

0.94

stad

0.88

lihood

0.88

ratulations

0.87

nesday

0.86

nton

0.85

pton

0.85

cheon

0.84

Activations Density 0.009%

mentions of the word "wich" in the text

references to specific geographic locations, particularly places ending in "wich."

No Comments

No Known Activations