INDEX

Explanations

references to influential people and their achievements or labeled identities

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

olutions

-0.65

 overload

-0.64

IPM

-0.61

 overfl

-0.60

 pharmacies

-0.58

 aggregate

-0.57

 heated

-0.57

 offsets

-0.57

 Markets

-0.57

 shifts

-0.56

POSITIVE LOGITS

 himself

1.05

 persona

1.03

 whom

1.03

Jr

0.90

who

0.89

 deceased

0.86

 Himself

0.86

 heir

0.83

 charismatic

0.83

whose

0.82

Activations Density 1.109%

references to influential people and their achievements or labeled identities

No Comments

No Known Activations