INDEX

Explanations

numeric comparisons

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

phrases that indicate duration or time spent in a position or condition

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.hook_resid_post

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_post

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

iquid

-0.73

ucker

-0.72

iru

-0.69

olt

-0.68

yout

-0.67

ailability

-0.67

ouse

-0.67

 constitu

-0.67

 exting

-0.67

Limit

-0.66

POSITIVE LOGITS

 nearly

1.05

 decades

1.04

 seventeen

1.00

 eight

1.00

 thirteen

0.99

 seven

0.98

 nine

0.98

 fourteen

0.96

 eleven

0.96

 sixteen

0.95

Activations Density 0.172%

numeric comparisons

phrases that indicate duration or time spent in a position or condition

No Comments

No Known Activations