INDEX

Explanations

names of people and their positions or titles

oai_token-act-pair · gpt-3.5-turbo

instances of quotations or statements attributed to different individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 7-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.7.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.7.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

rats

-0.68

cffffcc

-0.67

Kind

-0.65

inces

-0.64

OX

-0.61

bucks

-0.61

commit

-0.61

Results

-0.60

eways

-0.60

 calendar

-0.60

POSITIVE LOGITS

Jr

1.22

Sr

1.04

aka

0.97

 dean

0.92

 superintendent

0.89

 director

0.85

MD

0.84

QC

0.84

 Managing

0.84

 chairman

0.83

Activations Density 0.085%

names of people and their positions or titles

instances of quotations or statements attributed to different individuals

No Comments

No Known Activations