INDEX

Explanations

phrases related to politics and legislative activities

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

 civilisation

-0.80

tics

-0.78

 decomp

-0.78

zsche

-0.77

 superst

-0.75

DragonMagazine

-0.74

 attribution

-0.72

 explorers

-0.71

 experien

-0.71

 catalogue

-0.70

POSITIVE LOGITS

Senate

1.65

Republican

1.55

 Senate

1.51

Republicans

1.51

Democrats

1.46

 senate

1.42

 caucus

1.41

 Democrat

1.40

Senator

1.40

 senators

1.39

Activations Density 4.319%

phrases related to politics and legislative activities

No Comments

No Known Activations