INDEX

Explanations

words related to professions and titles, especially in the field of medicine and journalism

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

obin

-0.82

linked

-0.77

ENN

-0.74

hari

-0.73

KT

-0.73

amed

-0.71

mo

-0.70

nels

-0.69

lehem

-0.68

roup

-0.67

POSITIVE LOGITS

 finest

0.96

own

0.95

 sake

0.95

 newest

0.95

 maiden

0.91

 penchant

0.85

 inability

0.84

 insistence

0.84

 downfall

0.83

 biggest

0.82

Activations Density 0.447%

words related to professions and titles, especially in the field of medicine and journalism

No Comments

No Known Activations