INDEX

Explanations

phrases or sentences that indicate strong emphasis or comparison

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

axter

-0.65

ãĤ¿

-0.63

kefeller

-0.63

bath

-0.63

icides

-0.61

kai

-0.61

HCR

-0.61

iday

-0.60

tions

-0.58

undo

-0.58

POSITIVE LOGITS

 partially

0.78

 partly

0.78

 SOME

0.73

uner

0.69

 toler

0.68

 pretend

0.67

 temporarily

0.66

hap

0.65

 theoretically

0.65

lik

0.65

Activations Density 1.122%

phrases or sentences that indicate strong emphasis or comparison

No Comments

No Known Activations