INDEX

Explanations

Names and abbreviations related to locations and individuals, potentially from social media or news sources

oai_token-act-pair · gpt-3.5-turbo

specific location or identifiers related to geographic or fictional places

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ascript

-0.75

ilial

-0.73

 Qiao

-0.70

catentry

-0.66

theless

-0.65

Versions

-0.65

PsyNetMessage

-0.62

ruction

-0.61

ãĥ¼ãĥĨãĤ£

-0.60

advertisement

-0.59

POSITIVE LOGITS

EStreamFrame

1.08

 Lumpur

1.06

ikuman

0.94

EStream

0.91

bye

0.89

wagen

0.85

boxing

0.79

nik

0.75

enei

0.75

lyak

0.74

Activations Density 0.597%

Names and abbreviations related to locations and individuals, potentially from social media or news sources

specific location or identifiers related to geographic or fictional places

No Comments

No Known Activations