INDEX

Explanations

text sequences resembling computer programming code or commands

oai_token-act-pair · gpt-3.5-turbo

references to journalism and news-related themes

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

iga

-0.81

enaries

-0.75

gements

-0.68

ãĥ¼ãĥĨ

-0.66

 Tsukuyomi

-0.65

igue

-0.65

esis

-0.64

rast

-0.64

 Machina

-0.63

atial

-0.63

POSITIVE LOGITS

LOS

0.99

WASHINGTON

0.88

POS

0.81

PROV

0.81

LAB

0.81

SHARE

0.79

Researchers

0.78

Writing

0.77

Scientists

0.77

Published

0.75

Activations Density 0.636%

text sequences resembling computer programming code or commands

references to journalism and news-related themes

No Comments

No Known Activations