INDEX

Explanations

phrases indicating a high level of certainty or emphasis

oai_token-act-pair · gpt-3.5-turbo

phrases that express a degree of certainty or specificity

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

eworks

-0.80

iddles

-0.79

packages

-0.77

###

-0.76

TeX

-0.75

 thence

-0.75

usercontent

-0.75

>>>

-0.75

enders

-0.74

âĢ¢âĢ¢âĢ¢âĢ¢

-0.72

POSITIVE LOGITS

 amount

1.46

 extent

1.12

 percentage

1.11

 subset

1.07

 number

1.06

 degree

1.02

ties

0.95

 kind

0.94

 threshold

0.91

 proportion

0.84

Activations Density 0.033%

phrases indicating a high level of certainty or emphasis

phrases that express a degree of certainty or specificity

No Comments

No Known Activations