INDEX

Explanations

references to the news agency "Associated Press"

oai_token-act-pair · gpt-3.5-turbo

mentions of the Associated Press

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

gone

-0.92

orously

-0.85

abiding

-0.79

ggle

-0.76

gart

-0.76

itars

-0.74

steen

-0.74

perm

-0.71

vae

-0.71

vet

-0.69

POSITIVE LOGITS

 Newsp

0.99

 Colleg

0.90

 Press

0.88

 Association

0.83

 Arbit

0.80

 Tribune

0.80

 Agency

0.77

 Cooperative

0.76

Journal

0.76

 Corporation

0.75

Activations Density 0.007%

references to the news agency "Associated Press"

mentions of the Associated Press

No Comments

No Known Activations