INDEX

Explanations

words related to the "Star Trek" franchise

oai_token-act-pair · gpt-3.5-turbo

references to the "Star Trek" franchise

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

###

-0.81

 masked

-0.72

 suppressing

-0.66

 humid

-0.65

\'

-0.64

 execution

-0.64

 obsc

-0.63

ilies

-0.63

oad

-0.62

area

-0.61

POSITIVE LOGITS

 Trek

3.95

 Starfleet

1.80

Tre

1.60

 Klingon

1.56

 trek

1.53

 Spock

1.52

 Picard

1.38

 Voyager

1.32

 Vulcan

1.27

 Kling

1.25

Activations Density 0.020%

words related to the "Star Trek" franchise

references to the "Star Trek" franchise

No Comments

No Known Activations