INDEX

Explanations

phrases related to future actions or predictions

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 antioxid

-0.91

 portfolios

-0.66

ortment

-0.60

hoe

-0.60

 upgr

-0.60

 some

-0.58

redients

-0.58

BP

-0.58

ettlement

-0.56

rather

-0.56

POSITIVE LOGITS

theless

0.96

nor

0.93

rences

0.81

 dime

0.81

 dreamed

0.78

Initialized

0.73

before

0.72

 penny

0.70

 EVER

0.69

cation

0.69

Activations Density 0.156%

phrases related to future actions or predictions

No Comments

No Known Activations