INDEX

Explanations

references to specific television networks

oai_token-act-pair · gpt-3.5-turbo

mentions of major television networks, especially HBO

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

uation

-0.94

alion

-0.86

uart

-0.82

ional

-0.80

acle

-0.76

acular

-0.74

der

-0.73

cking

-0.73

letal

-0.73

ually

-0.72

POSITIVE LOGITS

BILL

0.70

HBO

0.70

GO

0.70

 Wire

0.69

OST

0.69

 Jinn

0.68

WAYS

0.67

IGHTS

0.67

 Host

0.66

otos

0.66

Activations Density 0.043%

references to specific television networks

mentions of major television networks, especially HBO

No Comments

No Known Activations