INDEX

Explanations

mentions of rankings or positions, particularly those starting with "Top" followed by a number

oai_token-act-pair · gpt-3.5-turbo

references to ranking lists, specifically "Top X" lists related to various categories

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.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

ufact

-0.81

 Gaul

-0.66

fir

-0.66

issance

-0.66

oresc

-0.65

BILITY

-0.64

 Advent

-0.63

Â¯Â¯

-0.63

 heightened

-0.63

 stead

-0.62

POSITIVE LOGITS

eka

1.12

Top

0.96

ographical

0.94

ography

0.92

most

0.85

ology

0.85

Boss

0.82

Top

0.82

mast

0.82

ICS

0.81

Activations Density 0.023%

mentions of rankings or positions, particularly those starting with "Top" followed by a number

references to ranking lists, specifically "Top X" lists related to various categories

No Comments

No Known Activations