INDEX

Explanations

verbs related to exploration, investigation, or revelation

oai_token-act-pair · gpt-3.5-turbo

instances of the word "discovered."

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

stay

-0.63

orting

-0.60

 tone

-0.59

voice

-0.59

depending

-0.59

regulation

-0.58

drive

-0.58

clinton

-0.57

VOL

-0.57

 fray

-0.57

POSITIVE LOGITS

 discovered

3.29

 unearthed

2.13

 uncovered

2.07

 discovers

1.95

 discover

1.93

 found

1.89

 detected

1.85

 noticed

1.79

 discovering

1.73

 discoveries

1.67

Activations Density 0.013%

verbs related to exploration, investigation, or revelation

instances of the word "discovered."

No Comments

No Known Activations