INDEX

Explanations

The neuron is looking for names or terms containing 'ene', potentially related to names, places, or chemical substances

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.8.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

ties

-0.86

ernaut

-0.86

istani

-0.79

icago

-0.76

RANT

-0.75

oons

-0.75

DOWN

-0.73

ricane

-0.73

uitous

-0.72

enegger

-0.72

POSITIVE LOGITS

uve

1.18

utics

1.07

utical

1.05

agles

1.05

alogy

1.01

 Wenger

0.99

bris

0.96

agle

0.95

quin

0.95

volent

0.94

Activations Density 0.027%

The neuron is looking for names or terms containing 'ene', potentially related to names, places, or chemical substances

No Comments

No Known Activations