INDEX

Explanations

instances where something has doubled

oai_token-act-pair · gpt-3.5-turbo

instances of the word "doubled" or its variations in the context of measured quantities

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Services

-0.88

utic

-0.72

uay

-0.72

 Condition

-0.71

ainment

-0.66

EMS

-0.65

ä½ľ

-0.65

çĦ

-0.65

fw

-0.64

Interstitial

-0.64

POSITIVE LOGITS

 doubled

0.88

 tripled

0.87

 doubling

0.78

ouble

0.75

 digit

0.73

ĸļ

0.72

 exponentially

0.71

 dipping

0.69

 overboard

0.67

 overtime

0.67

Activations Density 0.007%

instances where something has doubled

instances of the word "doubled" or its variations in the context of measured quantities

No Comments

No Known Activations