INDEX

Explanations

statements made by officials or military sources

oai_token-act-pair · gpt-3.5-turbo

phrases indicating reported speech or citations from sources

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

otin

-0.96

Measure

-0.72

 ILCS

-0.71

Oak

-0.70

Roberts

-0.70

Lot

-0.68

COR

-0.68

Newsletter

-0.68

Honest

-0.67

weed

-0.67

POSITIVE LOGITS

 afterwards

0.82

ibi

0.80

Downloadha

0.78

doms

0.78

YPG

0.76

 Mahm

0.75

ua

0.73

istani

0.72

orean

0.72

oÄŁ

0.71

Activations Density 0.197%

statements made by officials or military sources

phrases indicating reported speech or citations from sources

No Comments

No Known Activations