INDEX

Explanations

phrases indicating types of actions, events, or conditions

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

aceae

-0.71

 âĶľ

-0.68

agree

-0.66

 âĶľâĶĢâĶĢ

-0.64

lords

-0.63

ngth

-0.63

livious

-0.63

 somet

-0.62

grounds

-0.61

nown

-0.61

POSITIVE LOGITS

 moratorium

1.19

 boycott

1.13

 halt

1.06

 truce

0.91

 referendum

0.87

 barrage

0.85

 roundup

0.83

 dismissal

0.83

 demonstration

0.83

 crackdown

0.82

Activations Density 0.078%

phrases indicating types of actions, events, or conditions

No Comments

No Known Activations