INDEX

Explanations

terms related to authority or issuing commands

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

 istg

-0.75

bilt

-0.74

inth

-0.70

————

-0.66

bourne

-0.65

isphere

-0.63

âĢ¢âĢ¢âĢ¢âĢ¢

-0.62

olitics

-0.62

±

-0.60

ynski

-0.59

POSITIVE LOGITS

igraph

1.11

backs

1.06

 attention

0.96

0.95

 forth

0.88

 bluff

0.83

phas

0.82

oused

0.82

aghan

0.81

 upon

0.80

Activations Density 0.448%

terms related to authority or issuing commands

No Comments

No Known Activations