INDEX

Explanations

instances where someone sets up something new, like an office or a website

oai_token-act-pair · gpt-3.5-turbo

phrases related to establishing new entities or initiatives

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

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

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 embodiments

-0.74

iev

-0.72

 embodiment

-0.67

 understatement

-0.67

unknown

-0.64

enance

-0.64

 ions

-0.63

oft

-0.63

agree

-0.63

akis

-0.62

POSITIVE LOGITS

new

0.94

 foothold

0.80

 rapport

0.79

 rudimentary

0.78

 separate

0.75

 slew

0.75

 bunch

0.73

 dozen

0.73

 makeshift

0.72

 temporary

0.72

Activations Density 0.178%

instances where someone sets up something new, like an office or a website

phrases related to establishing new entities or initiatives

No Known Activations