INDEX

Explanations

keywords related to misinformation or inaccuracies

oai_token-act-pair · gpt-3.5-turbo

instances of high numerical values or scores in the context of action or compliance

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.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

 endeavour

-0.67

utical

-0.66

igi

-0.63

 equip

-0.62

earch

-0.61

 assignment

-0.61

 undermin

-0.61

 Shinra

-0.58

 Reincarnated

-0.58

 deceived

-0.58

POSITIVE LOGITS

ÂłÂłÂłÂł

0.99

ÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂłÂł

0.92

ÂłÂłÂłÂłÂłÂłÂłÂł

0.86

SPONSORED

0.85

ÂłÂłÂł

0.85

Newsletter

0.83

ÂłÂł

0.81

Commercial

0.80

Writing

0.77

Posted

0.76

Activations Density 1.001%

keywords related to misinformation or inaccuracies

instances of high numerical values or scores in the context of action or compliance

No Comments

No Known Activations