INDEX

Explanations

names, particularly first names of individuals

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

 Lamar

-0.76

 Newt

-0.68

 JosÃ©

-0.67

 Reverend

-0.67

 Buddy

-0.67

 Shaun

-0.66

 Enrique

-0.65

 Pixie

-0.64

 Dwight

-0.64

 Weird

-0.62

POSITIVE LOGITS

ovich

0.98

acci

0.95

otte

0.94

ovsky

0.94

opoulos

0.93

enstein

0.92

insky

0.92

shaw

0.92

atos

0.91

fman

0.91

Activations Density 0.053%

names, particularly first names of individuals

No Comments

No Known Activations