INDEX

Explanations

the word "bur" followed by a number indicating intensity or direction

oai_token-act-pair · gpt-3.5-turbo

references to burrowing behavior or characteristics related to burrows

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

é¾įå

-0.84

aunder

-0.78

uncture

-0.74

éĹĺ

-0.74

 Regulatory

-0.72

ACTED

-0.72

 Failure

-0.69

peak

-0.68

eanor

-0.67

FTWARE

-0.66

POSITIVE LOGITS

bur

1.25

rows

1.14

rito

1.10

rowing

1.01

gey

0.92

bur

0.92

rette

0.90

row

0.88

geons

0.86

geon

0.86

Activations Density 0.004%

the word "bur" followed by a number indicating intensity or direction

references to burrowing behavior or characteristics related to burrows

No Comments

No Known Activations