INDEX

Explanations

mentions of historical, religious, and technical terms

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

Catalog

-0.78

 istg

-0.74

igue

-0.71

ername

-0.70

/"

-0.69

ucci

-0.67

ashtra

-0.67

.''.

-0.66

.-

-0.66

way

-0.66

POSITIVE LOGITS

 these

0.94

 each

0.83

 those

0.82

 mankind

0.82

 humankind

0.79

the

0.79

our

0.78

 oneself

0.73

 this

0.72

 incumb

0.69

Activations Density 3.952%

mentions of historical, religious, and technical terms

No Comments

No Known Activations