INDEX

Explanations

locations or places

oai_token-act-pair · gpt-3.5-turbo

references to specific locations or geographic entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 stockp

-0.72

acca

-0.67

 rigged

-0.66

 recess

-0.64

pees

-0.63

byn

-0.62

thood

-0.61

aceae

-0.61

 hashing

-0.61

falls

-0.61

POSITIVE LOGITS

 Herald

1.34

 Observer

1.31

 Chronicle

1.22

 Gazette

1.22

 Tribune

1.21

 Inqu

1.19

 Times

1.18

Times

1.16

 Examiner

1.15

 Sentinel

1.13

Activations Density 0.088%

locations or places

references to specific locations or geographic entities

No Comments

No Known Activations