INDEX

Explanations

instances of the word "said"

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

theless

-0.73

ãĥİ

-0.73

 pend

-0.70

resent

-0.69

paralle

-0.68

thur

-0.67

 conflic

-0.67

MU

-0.67

pend

-0.65

earable

-0.63

POSITIVE LOGITS

 sarcast

0.95

 bluntly

0.90

 rhet

0.88

 referring

0.85

 emphatically

0.78

 aloud

0.77

 afterward

0.75

 diplom

0.74

 incred

0.74

 proudly

0.72

Activations Density 0.116%

instances of the word "said"

No Comments

No Known Activations