INDEX

Explanations

mentions of walls

oai_token-act-pair · gpt-3.5-turbo

references to barriers or obstructions, particularly walls

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

judicial

-0.66

 predictive

-0.60

 convenient

-0.58

 Coch

-0.58

Gene

-0.56

heny

-0.55

 pivotal

-0.55

ELS

-0.54

 skilled

-0.54

 practiced

-0.54

POSITIVE LOGITS

abies

1.17

wall

1.14

papers

1.00

paper

0.96

aby

0.94

stones

0.91

igans

0.81

odon

0.80

agos

0.79

 wallpaper

0.79

Activations Density 0.005%

mentions of walls

references to barriers or obstructions, particularly walls

No Comments

No Known Activations