INDEX

Explanations

phrases related to specific individuals' names and 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

 mascul

-0.63

 surg

-0.57

PAC

-0.55

 conce

-0.54

 skim

-0.54

 initials

-0.53

 reckoning

-0.53

 Converted

-0.52

 refrain

-0.52

 brace

-0.52

POSITIVE LOGITS

oop

0.84

ood

0.83

uese

0.83

oad

0.82

ainer

0.82

obia

0.81

andise

0.80

opsis

0.80

ilk

0.79

ainers

0.79

Activations Density 1.077%

phrases related to specific individuals' names and events

No Comments

No Known Activations