INDEX

Explanations

mention of physical structures like walls

oai_token-act-pair · gpt-3.5-turbo

references to walls

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

Gene

-0.88

Pacific

-0.73

Wild

-0.69

phrine

-0.68

 delegated

-0.67

lihood

-0.67

Eug

-0.66

ECK

-0.63

milo

-0.63

orean

-0.63

POSITIVE LOGITS

papers

1.22

abies

1.03

 walls

1.02

 wall

1.02

aby

1.00

buster

0.93

 thickness

0.89

paper

0.89

clock

0.89

 erected

0.87

Activations Density 0.027%

mention of physical structures like walls

references to walls

No Comments

No Known Activations