INDEX

Explanations

words related to opposition or conflict

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

ipel

-0.69

uilt

-0.67

enegger

-0.67

ibly

-0.62

ITED

-0.61

IOR

-0.61

insured

-0.61

ollen

-0.59

iven

-0.58

 affected

-0.58

POSITIVE LOGITS

 heels

0.88

 glove

0.74

toe

0.69

 pound

0.68

 knees

0.67

 tails

0.67

 bisc

0.66

quo

0.66

 roses

0.65

 oranges

0.64

Activations Density 6.529%

words related to opposition or conflict

No Comments

No Known Activations