INDEX

Explanations

uppercase words that might indicate system commands or technical instructions

oai_token-act-pair · gpt-3.5-turbo

occurrences of the abbreviation "OW" in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ãĤ¶

-0.87

ãĥ¼ãĥĨ

-0.83

ãĤ´ãĥ³

-0.70

ãĥ¤

-0.70

 Gupta

-0.66

kson

-0.66

 suspic

-0.65

ority

-0.65

 à¨

-0.63

 gratification

-0.63

POSITIVE LOGITS

orld

1.14

OW

1.06

DER

1.06

ARDS

1.05

LAN

1.04

atts

0.98

ARD

0.95

idth

0.92

lean

0.92

ERS

0.91

Activations Density 0.008%

uppercase words that might indicate system commands or technical instructions

occurrences of the abbreviation "OW" in various contexts

No Comments

No Known Activations