INDEX

Explanations

references to cheering or expressions of support and encouragement

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

ngth

-0.73

amation

-0.66

è£ħ

-0.63

nesota

-0.63

 fung

-0.58

ahime

-0.58

 consulted

-0.58

 Purg

-0.57

uture

-0.57

arin

-0.57

POSITIVE LOGITS

leaders

1.47

leader

1.44

leading

1.36

lead

1.19

fulness

1.04

 cheer

1.02

fully

0.95

wart

0.89

hovah

0.89

sticks

0.88

Activations Density 0.005%

references to cheering or expressions of support and encouragement

No Comments

No Known Activations