INDEX

Explanations

terms related to correction, adjustment, or fixing

oai_token-act-pair · gpt-3.5-turbo

terms related to the concept of rectification or correction

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

PLIED

-0.72

ARK

-0.71

UGH

-0.71

bang

-0.64

âĸ¬âĸ¬

-0.63

Fey

-0.63

Ø§ÙĦ

-0.62

EEP

-0.60

ESH

-0.60

WAYS

-0.59

POSITIVE LOGITS

 rect

1.19

angles

1.17

rect

1.12

angular

1.05

ified

1.01

uitous

0.98

rices

0.96

ifying

0.92

ifiable

0.92

itude

0.90

Activations Density 0.004%

terms related to correction, adjustment, or fixing

terms related to the concept of rectification or correction

No Comments

No Known Activations