INDEX

Explanations

references to flashlights or flashes of light

oai_token-act-pair · gpt-3.5-turbo

references to flash-related concepts or items

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Seym

-0.76

ally

-0.74

ufact

-0.72

 Dame

-0.70

ality

-0.69

romeda

-0.65

yip

-0.65

 Alban

-0.64

xus

-0.64

yss

-0.64

POSITIVE LOGITS

bang

1.16

lights

1.14

cards

1.13

points

1.12

point

1.11

light

1.10

fast

1.04

mob

0.98

trap

0.95

bul

0.94

Activations Density 0.028%

references to flashlights or flashes of light

references to flash-related concepts or items

No Comments

No Known Activations