INDEX

Explanations

large numerical values appearing in a technical context

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

vironment

-0.97

ĸļ

-0.71

 ãĤµãĥ¼ãĥĨãĤ£ãĥ¯ãĥ³

-0.70

 gradient

-0.66

 Jury

-0.63

 Rampage

-0.62

 vide

-0.62

 dime

-0.61

earch

-0.59

 preference

-0.58

POSITIVE LOGITS

upon

0.72

ridden

0.64

swick

0.62

Quotes

0.61

agall

0.61

export

0.60

 Shak

0.60

bart

0.59

eri

0.59

tips

0.58

Activations Density 1.778%

large numerical values appearing in a technical context

No Comments

No Known Activations