INDEX

Explanations

terms related to wires or wiring

oai_token-act-pair · gpt-3.5-turbo

mentions of "wire" and related terms, particularly in contexts involving surveillance or technology

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

 colonies

-0.68

 languages

-0.66

 grammar

-0.64

pag

-0.64

 errors

-0.63

 Sass

-0.62

 colony

-0.60

Nah

-0.60

 refreshing

-0.60

mon

-0.60

POSITIVE LOGITS

wire

4.77

Wire

2.77

wired

1.93

 Wire

1.48

 wire

1.08

wall

1.06

hack

1.01

Wheel

0.97

 wires

0.88

volt

0.86

Activations Density 0.003%

terms related to wires or wiring

mentions of "wire" and related terms, particularly in contexts involving surveillance or technology

No Comments

No Known Activations