INDEX

Explanations

references to gates or gate-related concepts

oai_token-act-pair · gpt-3.5-turbo

references to gates or barriers in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.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.4.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

ortium

-0.89

enegger

-0.81

lihood

-0.81

 Hots

-0.79

ensional

-0.67

issance

-0.65

inho

-0.64

ocker

-0.61

ulous

-0.61

 Norm

-0.60

POSITIVE LOGITS

keepers

1.31

keeper

1.27

 gates

1.15

ways

1.10

 gate

1.10

stones

1.00

posts

0.98

keeping

0.95

house

0.94

hole

0.91

Activations Density 0.007%

references to gates or gate-related concepts

references to gates or barriers in various contexts

No Comments

No Known Activations