INDEX

Explanations

phrases related to news articles or updates

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

schild

-0.66

ÄŁ

-0.65

Introduced

-0.61

ATTLE

-0.60

ynamic

-0.58

 Cummings

-0.56

ktop

-0.55

enegger

-0.55

ways

-0.54

 Daly

-0.53

POSITIVE LOGITS

ongyang

0.95

sylvania

0.90

ciation

0.85

asus

0.80

umatic

0.79

ĵĺ

0.76

merga

0.76

formance

0.74

cipled

0.73

uese

0.73

Activations Density 1.293%

phrases related to news articles or updates

No Comments

No Known Activations