INDEX

Explanations

titles or attributions from news sources

oai_token-act-pair · gpt-3.5-turbo

references to sources or attributions in the text

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

verages

-0.73

velop

-0.69

merce

-0.69

cknow

-0.67

 trillions

-0.66

ĸļ

-0.66

externalActionCode

-0.65

$$$$

-0.64

 numbering

-0.60

 sole

-0.60

POSITIVE LOGITS

usat

0.85

 VICE

0.78

related

0.74

âĢ¦]

0.71

 Carney

0.70

 POLITICO

0.69

...]

0.69

erous

0.69

amazon

0.65

 HuffPost

0.64

Activations Density 0.047%

titles or attributions from news sources

references to sources or attributions in the text

No Comments

No Known Activations