INDEX

Explanations

the word "fully"

oai_token-act-pair · gpt-3.5-turbo

mentions of a specific entity or brand

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

Wis

-0.70

 charts

-0.69

 chart

-0.67

 tips

-0.64

Pes

-0.64

 planners

-0.63

 meanings

-0.62

 circ

-0.61

 Centers

-0.61

 rulings

-0.61

POSITIVE LOGITS

ully

4.40

ULL

1.86

ull

1.21

ullivan

1.16

illy

1.14

usky

1.08

illet

1.05

uously

1.01

uther

1.00

rawler

0.98

Activations Density 0.015%

the word "fully"

mentions of a specific entity or brand

No Comments

No Known Activations