INDEX

Explanations

phrases related to political statements or opinions

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

ress

-0.68

arest

-0.66

andem

-0.65

Pont

-0.65

Tank

-0.64

oses

-0.63

aukee

-0.63

estern

-0.62

Guard

-0.62

Î±

-0.61

POSITIVE LOGITS

 although

1.13

"[

1.10

'[

0.85

 "...

0.85

 they

0.85

 "âĢ¦

0.84

soever

0.82

 there

0.81

 whilst

0.79

 while

0.77

Activations Density 0.537%

phrases related to political statements or opinions

No Comments

No Known Activations