INDEX

Explanations

contractions and negations, focusing on words like "don't" and "can't"

oai_token-act-pair · gpt-3.5-turbo

phrases expressing uncertainty or ignorance about knowledge or information

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

Not in Any Lists

No Comments

Negative Logits

lag

-0.68

 surges

-0.67

tein

-0.66

DAC

-0.63

ERG

-0.61

artifacts

-0.60

 Shock

-0.60

 helicop

-0.58

 Lamb

-0.56

 Asset

-0.56

POSITIVE LOGITS

 otherwise

0.91

 comprehend

0.79

 ordinarily

0.75

 approve

0.71

 understand

0.70

 understands

0.69

xa

0.68

 tolerate

0.68

ledge

0.67

 grasp

0.67

Activations Density 0.111%

contractions and negations, focusing on words like "don't" and "can't"

phrases expressing uncertainty or ignorance about knowledge or information

No Comments

No Known Activations