INDEX

Explanations

mentions of political leaders and legislative positions

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

kus

-0.72

eros

-0.72

arial

-0.71

uras

-0.69

 sled

-0.68

schild

-0.68

ourced

-0.67

urat

-0.67

ibaba

-0.65

erial

-0.62

POSITIVE LOGITS

 Newt

0.86

 Speaker

0.85

woman

0.84

 Boehner

0.82

person

0.77

 Gingrich

0.76

llan

0.76

 speaker

0.75

ower

0.73

 aide

0.72

Activations Density 0.006%

mentions of political leaders and legislative positions

No Comments

No Known Activations