INDEX

Explanations

mentions of political figures, particularly presidents and high-ranking officials

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

lamm

-0.68

estival

-0.68

fw

-0.67

ickle

-0.66

endi

-0.66

 radius

-0.66

ensitivity

-0.65

atum

-0.64

Sensor

-0.64

IFF

-0.63

POSITIVE LOGITS

 Yugoslavia

0.87

 Saddam

0.84

 Lyndon

0.81

 Colin

0.80

turned

0.79

 Edward

0.76

 Yugoslav

0.74

 Watergate

0.74

 Dennis

0.73

 Sheila

0.73

Activations Density 0.131%

mentions of political figures, particularly presidents and high-ranking officials

No Comments

No Known Activations