INDEX

Explanations

references to television programs or episodes airing on specific channels and dates

oai_token-act-pair · gpt-3.5-turbo

references to television programming and broadcasting

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 Templ

-0.65

 treasurer

-0.62

 Fraz

-0.61

 heirs

-0.58

 abstraction

-0.57

 Mead

-0.56

 Painter

-0.55

 Barron

-0.55

currency

-0.55

 manag

-0.55

POSITIVE LOGITS

 airing

1.12

 aired

1.02

 airs

0.94

 Tues

0.94

 tonight

0.94

episode

0.94

 Sundays

0.93

 episodes

0.90

live

0.90

 Mondays

0.88

Activations Density 0.058%

references to television programs or episodes airing on specific channels and dates

references to television programming and broadcasting

No Comments

No Known Activations