INDEX

Explanations

words related to broadcasting or airing content, such as 'televised' and 'livestream'

oai_token-act-pair · gpt-3.5-turbo

mentions of televised broadcasts or events

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

hart

-0.78

ateral

-0.70

une

-0.70

tta

-0.69

istries

-0.68

neg

-0.67

abet

-0.67

err

-0.66

rank

-0.65

upid

-0.65

POSITIVE LOGITS

 televised

1.10

 broadcasts

0.82

 announ

0.77

soDeliveryDate

0.75

 spectacle

0.74

é¾

0.73

 broadcast

0.72

 appearances

0.71

 millenn

0.70

 corrid

0.69

Activations Density 0.017%

words related to broadcasting or airing content, such as 'televised' and 'livestream'

mentions of televised broadcasts or events

No Comments

No Known Activations