INDEX

Explanations

words related to official instructions or commands

oai_token-act-pair · gpt-3.5-turbo

mentions of directives related to regulations or policies

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

ock

-0.75

 Blacks

-0.75

 Alto

-0.72

van

-0.72

 Argon

-0.71

 Sack

-0.67

riv

-0.66

OWS

-0.65

ikes

-0.65

ITE

-0.65

POSITIVE LOGITS

 directives

1.26

 directive

1.12

 Directive

0.93

 decree

0.90

 confir

0.90

 guiActiveUn

0.89

 directs

0.86

nod

0.84

 dictated

0.80

 nomine

0.80

Activations Density 0.008%

words related to official instructions or commands

mentions of directives related to regulations or policies

No Comments

No Known Activations