INDEX

Explanations

questions or statements ended in question marks

oai_token-act-pair · gpt-3.5-turbo

dialogue and questions expressed by characters

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

------------------------------------------------

-0.75

 endorsements

-0.66

 accompl

-0.63

 successes

-0.60

 victories

-0.59

 accord

-0.57

 acknowled

-0.57

fixes

-0.57

 surviving

-0.57

toc

-0.56

POSITIVE LOGITS

 asked

1.88

 asks

1.84

 inquired

1.74

 wondered

1.53

ask

1.48

 questions

1.47

 quer

1.43

 question

1.38

 questioned

1.38

 demanded

1.32

Activations Density 0.136%

questions or statements ended in question marks

dialogue and questions expressed by characters

No Comments

No Known Activations