INDEX

Explanations

references to the abbreviation "SH" in various contexts

oai_token-act-pair · gpt-4o-mini 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

heid

-0.91

naire

-0.85

leon

-0.76

ãĥ¼ãĥĨãĤ£

-0.73

tainment

-0.73

lia

-0.72

iform

-0.71

amide

-0.68

 agric

-0.67

sov

-0.66

POSITIVE LOGITS

IELD

1.44

IFT

1.30

ORTS

1.29

IPP

1.26

IPS

1.17

OCK

1.15

OTS

1.14

ORT

1.12

ARE

1.11

APE

1.11

Activations Density 0.013%

references to the abbreviation "SH" in various contexts

No Comments

No Known Activations