INDEX

Explanations

names of people along with their titles and locations

oai_token-act-pair · gpt-3.5-turbo

phrases related to organizational or regulatory actions

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Defense

-0.94

 favors

-0.94

 Hawai

-0.91

colored

-0.88

avored

-0.88

defense

-0.86

 Whitman

-0.86

caliber

-0.85

 favor

-0.84

Texas

-0.82

POSITIVE LOGITS

 Â£

1.46

 recognise

1.44

 apologise

1.42

 realise

1.40

 recognised

1.34

 analyse

1.33

 organising

1.31

 aluminium

1.31

 mould

1.31

 organise

1.31

Activations Density 2.132%

names of people along with their titles and locations

phrases related to organizational or regulatory actions

No Comments

No Known Activations