INDEX

Explanations

phrases related to news headlines

oai_token-act-pair · gpt-3.5-turbo

punctuation or symbols that mark the end of sections or sentences

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 comprom

-0.72

fee

-0.69

antage

-0.68

course

-0.68

 explan

-0.67

spr

-0.66

zu

-0.66

iod

-0.65

cf

-0.64

 stabil

-0.64

POSITIVE LOGITS

 Thousands

0.96

 Dozens

0.92

 Former

0.89

 Latest

0.89

 Hundreds

0.87

 WATCHED

0.87

 Investigators

0.87

Thousands

0.86

 Nearly

0.86

 Millions

0.84

Activations Density 0.019%

phrases related to news headlines

punctuation or symbols that mark the end of sections or sentences

No Comments

No Known Activations