INDEX

Explanations

mentions of specific publications or journalists

oai_token-act-pair · gpt-3.5-turbo

punctuation marks and formatting characters indicative of metadata or authorship in a document

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 whilst

-0.83

yip

-0.78

 secondly

-0.70

 eleph

-0.66

 eleven

-0.64

 util

-0.64

 flats

-0.63

senal

-0.63

 withd

-0.62

ilst

-0.61

POSITIVE LOGITS

 Caption

0.77

WASHINGTON

0.75

CLOSE

0.75

Consider

0.71

Editor

0.69

Politics

0.69

Correct

0.67

avage

0.66

 NAACP

0.65

Nearly

0.64

Activations Density 0.117%

mentions of specific publications or journalists

punctuation marks and formatting characters indicative of metadata or authorship in a document

No Comments

No Known Activations