INDEX

Explanations

claims that are unsupported or fabricated

oai_token-act-pair · gpt-3.5-turbo

terms related to false information and debunking claims

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

 actionGroup

-0.91

 mobility

-0.74

onding

-0.73

ktop

-0.72

rontal

-0.71

isites

-0.70

 hosp

-0.69

 Lumpur

-0.69

mediate

-0.68

itaire

-0.68

POSITIVE LOGITS

 assertions

1.27

 falsehood

1.27

 debunked

1.26

 untrue

1.16

 misinformation

1.14

False

1.13

 claims

1.12

 baseless

1.09

Claim

1.08

 debunk

1.07

Activations Density 0.378%

claims that are unsupported or fabricated

terms related to false information and debunking claims

No Comments

No Known Activations