INDEX

Explanations

misspelled words and typos

oai_token-act-pair · gpt-3.5-turbo

questions and user engagement indicators

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 ÂŃ

-0.76

AMY

-0.76

)—

-0.74

ARDIS

-0.71

 principals

-0.70

—-

-0.69

âĢĲ

-0.67

––

-0.67

)].

-0.66

emporary

-0.66

POSITIVE LOGITS

 Protoss

1.27

 Moder

1.25

 Zerg

1.21

 Profile

1.16

Poll

1.13

 Starcraft

1.11

 Liqu

1.09

 Slayer

1.06

 Administrator

1.05

MLG

1.04

Activations Density 0.100%

misspelled words and typos

questions and user engagement indicators

No Comments

No Known Activations