INDEX

Explanations

words related to news headlines or events

oai_token-act-pair · gpt-3.5-turbo

phrases indicating the closing of certain events or sections

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

 disgu

-0.86

hid

-0.83

beh

-0.79

ifice

-0.77

 dece

-0.77

ologically

-0.74

 outwe

-0.73

gery

-0.73

 ostr

-0.73

 rebell

-0.72

POSITIVE LOGITS

CLOSE

1.09

 Pacers

0.94

 Officials

0.91

 Researchers

0.90

 Below

0.88

 Hundreds

0.87

 Expand

0.82

 Here

0.82

UTERS

0.81

 Residents

0.81

Activations Density 0.018%

words related to news headlines or events

phrases indicating the closing of certain events or sections

No Comments

No Known Activations