INDEX

Explanations

proper nouns, particularly names of people and characters

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

pen

-0.81

 Robo

-0.80

ograph

-0.76

 captcha

-0.73

 simulator

-0.72

 Aero

-0.70

ãģ®å®

-0.69

 cardio

-0.69

 pens

-0.69

 Robots

-0.69

POSITIVE LOGITS

anda

1.11

ands

1.09

ande

1.03

AND

0.99

und

0.98

anding

0.92

hent

0.92

rand

0.88

alle

0.87

andan

0.87

Activations Density 0.089%

proper nouns, particularly names of people and characters

No Comments

No Known Activations