INDEX

Explanations

references to events or statements being televised

oai_token-act-pair · gpt-3.5-turbo

references to televised content or programming

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

hart

-0.72

istries

-0.71

uv

-0.66

ne

-0.65

une

-0.65

lua

-0.64

upon

-0.64

tta

-0.64

owder

-0.64

abiding

-0.63

POSITIVE LOGITS

 televised

1.11

 corrid

0.85

 broadcasts

0.84

soDeliveryDate

0.80

 announ

0.80

ÃįÃį

0.79

 broadcast

0.75

 EDITION

0.75

 satell

0.75

 WATCHED

0.74

Activations Density 0.009%

references to events or statements being televised

references to televised content or programming

No Comments

No Known Activations