INDEX

Explanations

names associated with high-profile legal cases and incidents involving police violence

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

ahime

-1.11

iculture

-0.81

road

-0.81

 Jagu

-0.78

ships

-0.77

 relegation

-0.74

roads

-0.72

eries

-0.72

 fielding

-0.69

rawdownloadcloneembedreportprint

-0.69

POSITIVE LOGITS

 Tsarnaev

0.89

 Snowden

0.89

ghazi

0.87

 assailant

0.86

FBI

0.84

 Saddam

0.81

 innocent

0.81

 gunman

0.80

«

0.80

 Laden

0.80

Activations Density 0.124%

names associated with high-profile legal cases and incidents involving police violence

No Comments

No Known Activations