INDEX

Explanations

references to physical doors

oai_token-act-pair · gpt-3.5-turbo

references to doors and door-related concepts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 wast

-0.71

POW

-0.70

 understatement

-0.61

 Lennon

-0.60

 includ

-0.59

Pow

-0.58

 proven

-0.58

UC

-0.58

 Sold

-0.58

ast

-0.58

POSITIVE LOGITS

door

4.79

doors

2.91

 Door

1.87

 door

1.85

 doors

1.68

 Doors

1.20

 doorway

1.19

trap

1.09

stairs

1.04

floor

1.02

Activations Density 0.008%

references to physical doors

references to doors and door-related concepts

No Comments

No Known Activations