INDEX

Explanations

adjectives and their variants

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

NSA

-0.82

employ

-0.79

REP

-0.70

MRI

-0.67

reports

-0.67

GV

-0.66

esson

-0.66

 Collider

-0.66

NX

-0.64

paid

-0.64

POSITIVE LOGITS

 paradise

0.90

 footh

0.84

 soils

0.83

 slopes

0.83

 greens

0.82

 slope

0.82

 plains

0.82

 areas

0.82

 substr

0.82

 gardens

0.80

Activations Density 0.035%

adjectives and their variants

No Comments

No Known Activations