INDEX

Explanations

commands and instructions in an unknown language

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

Flash

-0.65

 braces

-0.60

 shuffle

-0.59

WARE

-0.59

IUM

-0.58

Bloom

-0.57

 prematurely

-0.57

bucks

-0.56

 Advertisement

-0.56

 nerves

-0.56

POSITIVE LOGITS

©¶æ¥µ

0.85

arent

0.77

ĸļ

0.77

onde

0.77

ont

0.76

£ı

0.74

Ã©

0.73

otent

0.73

 alle

0.73

usalem

0.72

Activations Density 0.108%

commands and instructions in an unknown language

No Comments

No Known Activations