INDEX

Explanations

terms related to specific measurements or quantities

oai_token-act-pair · gpt-3.5-turbo

references to specific numerical quantities or statistics

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

QC

-0.74

 sche

-0.64

 scam

-0.59

 spam

-0.58

":"/

-0.58

 bump

-0.57

rad

-0.57

 boss

-0.55

uler

-0.55

-0.54

POSITIVE LOGITS

 Twenty

3.14

 Thirty

2.80

 Forty

2.63

Twenty

2.57

 Fifty

2.50

Thirty

2.30

ixty

2.20

Fif

2.05

 Sevent

1.91

 Twelve

1.71

Activations Density 0.022%

terms related to specific measurements or quantities

references to specific numerical quantities or statistics

No Comments

No Known Activations