INDEX

Explanations

references to alternative or additional items and concepts

oai_token-act-pair · gpt-4o-mini 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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

uckle

-0.70

oming

-0.69

iasis

-0.68

ensibly

-0.65

atism

-0.65

ONEY

-0.64

rought

-0.64

resa

-0.64

olver

-0.62

uffer

-0.62

POSITIVE LOGITS

 countries

1.30

 continents

1.29

worldly

1.25

 contexts

1.11

 Countries

1.09

 locations

1.09

 regions

1.06

 jurisdictions

1.03

 directions

0.99

 languages

0.98

Activations Density 0.079%

references to alternative or additional items and concepts

No Comments

No Known Activations