INDEX

Explanations

technical terms and concepts related to advanced technology and scientific research

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

 regulars

-0.80

 rotten

-0.78

#$#$

-0.77

 undone

-0.74

ggies

-0.73

natureconservancy

-0.73

 etiquette

-0.71

 volunt

-0.70

 innocence

-0.70

 excuses

-0.70

POSITIVE LOGITS

 Imaging

1.31

 imaging

1.22

 infrared

1.18

 Radiation

1.14

 Collider

1.10

imeter

1.10

 electromagnetic

1.09

 Sensor

1.09

 Detection

1.06

ensing

1.06

Activations Density 0.175%

technical terms and concepts related to advanced technology and scientific research

No Comments

No Known Activations