INDEX

Explanations

phrases indicating negation or denial

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

ItemThumbnailImage

-0.72

 selves

-0.70

 throats

-0.66

interstitial

-0.65

 Monitor

-0.63

stood

-0.61

 Tours

-0.61

Mo

-0.58

 chambers

-0.58

 arsen

-0.57

POSITIVE LOGITS

 necessarily

1.29

 advisable

0.92

 suffice

0.88

 uncommon

0.87

necess

0.86

 condone

0.86

icable

0.82

 feasible

0.82

eworthy

0.82

icably

0.81

Activations Density 0.094%

phrases indicating negation or denial

No Comments

No Known Activations