INDEX

Explanations

phrases related to logging or signing into accounts

oai_token-act-pair · gpt-3.5-turbo

references to logging and mapping data

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 superv

-0.71

 wartime

-0.68

vic

-0.67

ext

-0.67

 MacArthur

-0.65

 rage

-0.64

é

-0.64

 paternity

-0.63

 supporting

-0.63

 supers

-0.63

POSITIVE LOGITS

Log

3.01

Map

1.65

Mill

1.43

Mom

1.34

Organ

1.33

Magic

1.33

Wiki

1.25

Rule

1.24

Soft

1.24

Hide

1.23

Activations Density 0.041%

phrases related to logging or signing into accounts

references to logging and mapping data

No Comments

No Known Activations