INDEX

Explanations

articles or texts beginning with the word "Speaking" and followed by a location or person the speech is attributed to

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Speaking" in the text

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

ascade

-0.71

aden

-0.71

Ã©t

-0.69

\/\/

-0.69

jer

-0.69

=~

-0.68

involved

-0.68

gradation

-0.64

origin

-0.63

————

-0.63

POSITIVE LOGITS

 engagements

0.81

 aloud

0.77

 volumes

0.70

 spoke

0.70

gow

0.68

 apart

0.68

 McCartney

0.66

 speaking

0.66

 Languages

0.66

ificantly

0.65

Activations Density 0.012%

articles or texts beginning with the word "Speaking" and followed by a location or person the speech is attributed to

instances of the word "Speaking" in the text

No Comments

No Known Activations