INDEX

Explanations

words related to news reporting, specifically focusing on specific organizations and types of investigations

oai_token-act-pair · gpt-3.5-turbo

references to specific media reports and official investigations

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 magnets

-0.87

åĭ

-0.84

ELF

-0.78

eg

-0.77

cffff

-0.76

icks

-0.76

æ©Ł

-0.76

Els

-0.76

 allerg

-0.75

 Brus

-0.75

POSITIVE LOGITS

project

0.94

 Project

0.92

 project

0.90

 survey

0.89

ertation

0.85

 campaign

0.84

 investigation

0.83

 episode

0.83

 retrospective

0.82

APTER

0.81

Activations Density 0.352%

words related to news reporting, specifically focusing on specific organizations and types of investigations

references to specific media reports and official investigations

No Comments

No Known Activations