INDEX

Explanations

instances of the word "some" indicating quantities or unspecified amounts

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

istan

-0.67

atches

-0.67

meet

-0.66

ared

-0.64

gon

-0.64

ipples

-0.64

uckle

-0.64

raid

-0.63

eds

-0.62

Dub

-0.61

POSITIVE LOGITS

 semblance

1.36

body

1.03

place

1.02

ones

0.97

 sort

0.96

how

0.91

 kind

0.90

 indication

0.87

 degree

0.87

 unspecified

0.83

Activations Density 0.072%

instances of the word "some" indicating quantities or unspecified amounts

No Comments

No Known Activations