INDEX

Explanations

phrases starting with "are" and providing information or explanation

oai_token-act-pair · gpt-3.5-turbo

references to various entities or concepts in a factual context

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

west

-0.87

\\\\\\\\

-0.72

rower

-0.71

culosis

-0.71

Edit

-0.70

luaj

-0.70

dt

-0.69

hoff

-0.67

ennis

-0.66

edin

-0.66

POSITIVE LOGITS

 inherently

1.10

 notoriously

1.09

 prolifer

1.06

 interchangeable

1.05

 typically

1.03

 plentiful

1.03

 designed

1.03

 commonplace

1.02

 everywhere

0.99

 relics

0.99

Activations Density 0.274%

phrases starting with "are" and providing information or explanation

references to various entities or concepts in a factual context

No Comments

No Known Activations