INDEX

Explanations

personal possessive pronouns and possessive phrases

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

mers

-0.73

olid

-0.72

olic

-0.72

MER

-0.71

mes

-0.68

ammy

-0.66

mer

-0.66

rought

-0.64

obar

-0.64

Mub

-0.63

POSITIVE LOGITS

 accord

0.84

 selves

0.74

Cloud

0.72

 personal

0.69

 merits

0.69

 identifiable

0.69

selves

0.67

 Samoa

0.65

 affairs

0.65

 unique

0.63

Activations Density 3.257%

personal possessive pronouns and possessive phrases

No Comments

No Known Activations