INDEX

Explanations

phrases related to consequences of actions or delays

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

atari

-0.71

anon

-0.71

 Sheep

-0.69

=-=-=-=-=-=-=-=-

-0.63

ille

-0.63

arest

-0.62

 Seymour

-0.61

ivas

-0.60

arro

-0.60

aroo

-0.60

POSITIVE LOGITS

 diligence

1.15

giving

0.91

lling

0.78

itations

0.73

 dilig

0.72

 cancell

0.70

 solely

0.69

brance

0.69

itiz

0.66

iments

0.64

Activations Density 1.086%

phrases related to consequences of actions or delays

No Comments

No Known Activations