INDEX

Explanations

terms related to political swing states

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

sis

-0.73

 Lauder

-0.70

ABE

-0.69

xia

-0.69

cies

-0.69

 Seym

-0.68

udes

-0.68

src

-0.67

å£«

-0.66

resp

-0.64

POSITIVE LOGITS

 swinging

0.84

 swing

0.82

mill

0.80

shape

0.78

erman

0.77

halla

0.77

 pend

0.77

erer

0.76

grass

0.75

ers

0.73

Activations Density 0.014%

terms related to political swing states

No Comments

No Known Activations