INDEX

Explanations

phrases indicating publication or event timelines

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

ascript

-0.70

escape

-0.67

$$$$

-0.66

 honestly

-0.65

just

-0.63

ecause

-0.63

utations

-0.62

rams

-0.62

ollar

-0.61

mini

-0.60

POSITIVE LOGITS

 HuffPost

0.96

 Dangerous

0.76

 NEWS

0.70

 POLITICO

0.69

 Berk

0.67

 Alert

0.67

 Interest

0.66

 Forbes

0.65

 Heads

0.63

gov

0.63

Activations Density 0.025%

phrases indicating publication or event timelines

No Comments

No Known Activations