INDEX

Explanations

verbs or nouns related to actions, behavior, or events

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

ixed

-0.74

 Lank

-0.71

fer

-0.68

ļéĨĴ

-0.67

bid

-0.67

skinned

-0.67

ickets

-0.66

arton

-0.64

lined

-0.63

moon

-0.62

POSITIVE LOGITS

uated

1.17

uate

1.15

ives

1.12

uary

1.05

uating

1.05

ional

1.02

ual

1.00

uations

0.99

ivity

0.97

ivism

0.97

Activations Density 2.097%

verbs or nouns related to actions, behavior, or events

No Comments

No Known Activations