INDEX

Explanations

phrases related to reporting or presenting new information

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_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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

Entry

-0.62

maxwell

-0.59

enei

-0.59

itarian

-0.59

idth

-0.57

pmwiki

-0.56

ities

-0.55

tein

-0.54

 Metatron

-0.54

atever

-0.54

POSITIVE LOGITS

 earlier

0.95

 beforehand

0.95

 during

0.89

 yesterday

0.81

 prior

0.79

 before

0.77

 BEFORE

0.77

 throughout

0.75

 shortly

0.72

 previously

0.71

Activations Density 0.182%

phrases related to reporting or presenting new information

No Comments

No Known Activations