INDEX

Explanations

phrases indicating risk and implications of actions on individuals or groups

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

 refuses

-0.63

 uttered

-0.62

ulously

-0.61

nce

-0.61

 motions

-0.61

leys

-0.60

 forbids

-0.60

naires

-0.59

gans

-0.58

 lasts

-0.58

POSITIVE LOGITS

 jeopardy

1.00

pmwiki

0.84

 peril

0.83

ãĤ´ãĥ³

0.78

ãĥ¯ãĥ³

0.76

advant

0.73

 unwelcome

0.72

 uncomfortable

0.71

scape

0.70

ãĤ§

0.68

Activations Density 0.064%

phrases indicating risk and implications of actions on individuals or groups

No Comments

No Known Activations