INDEX

Explanations

questions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Question" followed by a numerical value or reference

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

rites

-0.83

urses

-0.77

ufact

-0.73

 corrid

-0.71

tsky

-0.69

orpor

-0.67

ntil

-0.65

olia

-0.65

yss

-0.65

mobi

-0.63

POSITIVE LOGITS

naires

1.70

naire

1.49

 answered

1.03

mark

0.93

 mark

0.93

 posed

0.90

answer

0.88

Ans

0.87

 asked

0.87

 unanswered

0.86

Activations Density 0.044%

questions

instances of the word "Question" followed by a numerical value or reference

No Comments

No Known Activations