INDEX

Explanations

keywords related to buildings or structures

oai_token-act-pair · gpt-3.5-turbo

references to various buildings, particularly a specific building associated with the World Trade Center

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

sha

-0.82

romeda

-0.82

CHAT

-0.81

xon

-0.78

nir

-0.74

ican

-0.73

imum

-0.72

Sharp

-0.72

phal

-0.71

sweet

-0.70

POSITIVE LOGITS

 blocks

1.10

 occupancy

1.02

 occupants

1.00

raper

0.97

 renovation

0.96

 renovations

0.95

Constructed

0.89

 demolition

0.88

 constructed

0.88

 erected

0.88

Activations Density 0.052%

keywords related to buildings or structures

references to various buildings, particularly a specific building associated with the World Trade Center

No Comments

No Known Activations