INDEX

Explanations

mentions of things being reliable

oai_token-act-pair · gpt-3.5-turbo

instances of the word "reliable."

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ften

-0.93

owitz

-0.89

horn

-0.84

ifling

-0.80

thus

-0.80

hunt

-0.78

ogenesis

-0.78

thur

-0.75

pper

-0.75

ony

-0.74

POSITIVE LOGITS

 reliable

1.13

 reliability

1.12

 unreliable

0.96

 estim

0.89

iability

0.89

 conclud

0.87

 trustworthy

0.87

iable

0.86

 narrator

0.84

 intervals

0.83

Activations Density 0.013%

mentions of things being reliable

instances of the word "reliable."

No Comments

No Known Activations