INDEX

Explanations

words related to elections, political victories, and political events

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

 advising

-0.63

senal

-0.62

bian

-0.60

rouch

-0.59

OSE

-0.59

rete

-0.58

 forth

-0.57

 filler

-0.57

appa

-0.57

Factor

-0.57

POSITIVE LOGITS

now

0.98

nings

0.96

 trophies

0.83

 hearts

0.81

ipeg

0.80

 victories

0.78

cest

0.78

ced

0.78

't

0.77

throp

0.76

Activations Density 4.313%

words related to elections, political victories, and political events

No Comments

No Known Activations