INDEX

Explanations

mentions of politicians 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_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

viation

-0.76

eric

-0.68

ruary

-0.66

lag

-0.63

 ISBN

-0.63

ibility

-0.63

estones

-0.62

ulia

-0.61

tion

-0.60

acl

-0.59

POSITIVE LOGITS

 Kavanaugh

0.85

Senate

0.73

 Dems

0.66

Ħ¢

0.65

Ford

0.62

%]

0.62

avanaugh

0.62

 accuser

0.61

').

0.61

asse

0.60

Activations Density 0.518%

mentions of politicians and political events

No Comments

No Known Activations