INDEX

Explanations

phrases related to disbelief or rejection of claims

oai_token-act-pair · gpt-3.5-turbo

terms related to claims that are considered invalid, unsupported, or lacking credibility

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

udeau

-0.83

illes

-0.81

hov

-0.79

odon

-0.73

 FranÃ§

-0.71

anse

-0.71

oise

-0.71

eeper

-0.70

toc

-0.69

irie

-0.69

POSITIVE LOGITS

 baseless

1.03

 unfounded

1.01

False

0.91

 allegations

0.90

 accusations

0.88

Rum

0.85

ãĥ¥

0.84

 false

0.83

 allegation

0.82

 theories

0.80

Activations Density 0.022%

phrases related to disbelief or rejection of claims

terms related to claims that are considered invalid, unsupported, or lacking credibility

No Comments

No Known Activations