INDEX

Explanations

the word "whenever" indicating situations or conditions that recur or are conditional

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

emetery

-0.71

pour

-0.67

arling

-0.66

itan

-0.65

ocker

-0.64

uton

-0.64

ettle

-0.64

heimer

-0.64

athan

-0.63

%);

-0.63

POSITIVE LOGITS

soever

1.31

 possible

1.09

 imaginable

0.99

 practicable

0.97

 feasible

0.92

 else

0.90

 conceivable

0.89

 somebody

0.85

 someone

0.82

wcsstore

0.79

Activations Density 0.032%

the word "whenever" indicating situations or conditions that recur or are conditional

No Comments

No Known Activations