INDEX

Explanations

numerical quantities and relationships

oai_token-act-pair · gpt-3.5-turbo

quantitative references or counts related to entities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

Mobil

-0.63

 ÂŃ

-0.61

—"

-0.56

pping

-0.55

emaker

-0.55

assy

-0.54

Outside

-0.53

uary

-0.52

polit

-0.51

vertising

-0.51

POSITIVE LOGITS

 embodiments

0.82

 subreddits

0.75

 pics

0.74

 poems

0.73

 quotations

0.73

 queens

0.71

 verbs

0.67

 translations

0.67

 coats

0.67

pointers

0.66

Activations Density 0.319%

numerical quantities and relationships

quantitative references or counts related to entities

No Comments

No Known Activations