INDEX

Explanations

words or phrases related to investigation or inquiry processes

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 numbered

-0.71

 Magnus

-0.67

abouts

-0.66

 structured

-0.66

 Sparkle

-0.64

è£

-0.63

ãĥ¯ãĥ³

-0.62

 Peer

-0.61

 Somerset

-0.60

 grit

-0.59

POSITIVE LOGITS

gered

0.97

raphic

0.95

iosity

0.94

uana

0.92

dict

0.90

raphics

0.88

nation

0.87

uation

0.87

uration

0.86

uously

0.84

Activations Density 0.006%

words or phrases related to investigation or inquiry processes

No Comments

No Known Activations

words or phrases related to investigation or inquiry processes

No Comments

No Known Activations