INDEX

Explanations

words related to political events and entities

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

idious

-0.80

iblical

-0.68

Downloadha

-0.68

 oneself

-0.67

yss

-0.67

inarily

-0.65

itious

-0.64

abolic

-0.63

amsung

-0.60

-0.59

POSITIVE LOGITS

 Orchestra

0.82

 Library

0.78

eworks

0.77

 Association

0.76

 Prix

0.75

mbudsman

0.75

lance

0.74

abulary

0.73

itent

0.73

atorium

0.71

Activations Density 6.728%

words related to political events and entities

No Comments

No Known Activations