INDEX

Explanations

phrases or sentences introducing additional information or details

oai_token-act-pair · gpt-3.5-turbo

references to additional information or topics

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 Consolid

-0.77

CHAT

-0.73

 Jinn

-0.69

 Inher

-0.68

 Peaks

-0.67

 Bers

-0.67

 Kazakh

-0.65

 Glover

-0.65

 Story

-0.63

Kas

-0.62

POSITIVE LOGITS

 than

1.14

than

1.06

efficient

0.87

fficient

0.79

ptive

0.79

ensed

0.78

ocations

0.77

erous

0.75

culated

0.74

venient

0.73

Activations Density 0.723%

phrases or sentences introducing additional information or details

references to additional information or topics

No Comments

No Known Activations