INDEX

Explanations

references to governance and authority figures

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

 delet

-0.77

oops

-0.69

aughtered

-0.69

 rumors

-0.62

eatures

-0.60

rontal

-0.60

rollment

-0.60

 rumor

-0.59

issan

-0.58

venient

-0.58

POSITIVE LOGITS

ankind

0.81

ateurs

0.77

 outweigh

0.75

 whilst

0.73

cients

0.69

 versus

0.68

 paramount

0.66

ĸļ

0.66

ãĢĤ

0.65

 amidst

0.63

Activations Density 0.315%

references to governance and authority figures

No Comments

No Known Activations