INDEX

Explanations

quotes from people speaking with high confidence or authority

oai_token-act-pair · gpt-3.5-turbo

direct quotes or reported speech

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

Torrent

-0.70

 rotting

-0.70

wings

-0.69

steam

-0.69

ondon

-0.68

 ILCS

-0.66

maps

-0.66

conservancy

-0.66

itect

-0.65

soDeliveryDate

-0.65

POSITIVE LOGITS

 interviewer

1.16

 rhet

1.11

Asked

1.10

 sarcast

1.06

 moderator

1.06

 applause

1.06

 interrupted

1.05

 Asked

1.04

 laughter

1.02

 chuck

1.01

Activations Density 0.507%

quotes from people speaking with high confidence or authority

direct quotes or reported speech

No Comments

No Known Activations