INDEX

Explanations

words related to location or position

oai_token-act-pair · gpt-3.5-turbo

terms related to something being "lying" or "underlying."

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

arta

-0.72

ANA

-0.66

èª

-0.66

 sped

-0.59

aba

-0.58

fruit

-0.57

 bead

-0.57

à¦

-0.57

sha

-0.56

 reconstruction

-0.55

POSITIVE LOGITS

lying

1.06

redients

0.78

cium

0.76

ratulations

0.76

utenant

0.75

tons

0.72

uctor

0.69

aeda

0.69

 Input

0.69

gaard

0.69

Activations Density 0.005%

words related to location or position

terms related to something being "lying" or "underlying."

No Comments

No Known Activations