INDEX

Explanations

references to electrons or words related to electronic processes

oai_token-act-pair · gpt-3.5-turbo

references to electors or electoral processes

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

SAN

-0.75

rama

-0.71

 seasonal

-0.68

 zone

-0.68

 penalty

-0.67

wrapper

-0.66

 fiction

-0.65

 flush

-0.65

ictions

-0.64

 slump

-0.63

POSITIVE LOGITS

 Elect

4.09

Elect

2.62

elect

2.12

 elect

2.12

 ELECT

2.07

 Electro

1.80

 electron

1.66

 electors

1.62

 elector

1.44

 electr

1.34

Activations Density 0.015%

references to electrons or words related to electronic processes

references to electors or electoral processes

No Comments

No Known Activations