INDEX

Explanations

instances of the word "mark"

oai_token-act-pair · gpt-3.5-turbo

references to numerical ratings or scores

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

ILLE

-0.90

abama

-0.74

 traged

-0.71

rouch

-0.69

anooga

-0.69

odka

-0.67

Kut

-0.66

urus

-0.66

agan

-0.65

ctory

-0.63

POSITIVE LOGITS

manship

1.46

downs

1.09

down

1.09

eters

1.05

 Twain

0.97

ups

0.96

eting

0.89

emark

0.86

lights

0.84

boards

0.81

Activations Density 0.048%

instances of the word "mark"

references to numerical ratings or scores

No Comments

No Known Activations