INDEX

Explanations

locations or properties described with specific square footage values

oai_token-act-pair · gpt-3.5-turbo

negative numerical values or measurements

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.8.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.8.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

 Creator

-0.72

 Rica

-0.70

 Scrib

-0.69

 poets

-0.67

 Twain

-0.67

 Defenders

-0.66

 phil

-0.66

 Silk

-0.65

 moderation

-0.65

 Prometheus

-0.65

POSITIVE LOGITS

pound

1.27

plus

1.22

capacity

1.11

derived

1.07

scale

1.07

year

1.07

page

1.06

mile

1.06

degree

1.05

volt

1.05

Activations Density 0.079%

locations or properties described with specific square footage values

negative numerical values or measurements

No Comments

No Known Activations