INDEX

Explanations

answers to questions

oai_token-act-pair · gpt-3.5-turbo

mentions of questions and answers

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

rites

-0.70

ected

-0.67

ufact

-0.62

orpor

-0.59

mil

-0.58

joining

-0.57

agically

-0.57

 corrid

-0.56

oiler

-0.56

uj

-0.55

POSITIVE LOGITS

naires

1.67

naire

1.52

 answered

1.22

 posed

1.20

 asked

1.15

answered

1.11

 answ

1.09

Answer

1.05

iddles

0.99

answer

0.99

Activations Density 0.069%

answers to questions

mentions of questions and answers

No Comments

No Known Activations