INDEX

Explanations

dates in a specific format

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

 certs

-0.79

cffffcc

-0.74

netflix

-0.68

enegger

-0.65

soDeliveryDate

-0.64

 Rapp

-0.63

oster

-0.61

ographic

-0.61

 Blumenthal

-0.60

UTH

-0.59

POSITIVE LOGITS

 conclud

0.84

 upset

0.77

goal

0.75

ãĥ¼ãĥĨãĤ£

0.74

comings

0.70

venge

0.65

 Highlights

0.63

 halftime

0.63

 territ

0.62

win

0.62

Activations Density 0.748%

dates in a specific format

No Comments

No Known Activations