INDEX

Explanations

repeated uses of the word "he."

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

 GEAR

-0.76

 Gems

-0.71

 Interest

-0.70

 Voltage

-0.68

 Engineers

-0.65

 reinforcement

-0.60

 Grid

-0.60

 Nichols

-0.59

 affiliate

-0.57

 Allied

-0.56

POSITIVE LOGITS

zbollah

1.36

aven

1.30

brew

1.19

ather

1.18

avier

1.17

pherd

1.14

gemony

1.13

ldon

1.11

aling

1.10

uristic

1.09

Activations Density 0.044%

repeated uses of the word "he."

No Comments

No Known Activations