INDEX

Explanations

questions

oai_token-act-pair · gpt-3.5-turbo

instances of the word "Question" and related query formats

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

asters

-0.82

 strokes

-0.69

 coales

-0.69

aper

-0.67

aped

-0.66

 decorations

-0.65

 flock

-0.65

syn

-0.65

 choke

-0.63

aster

-0.63

POSITIVE LOGITS

 Question

3.76

Question

3.17

question

2.22

 Questions

2.11

 Answer

1.89

 QUEST

1.86

Answer

1.64

 question

1.43

 Debate

1.39

Questions

1.38

Activations Density 0.035%

questions

instances of the word "Question" and related query formats

No Comments

No Known Activations