INDEX

Explanations

instructions or steps related to organizing information and tasks effectively

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 trainer

-0.65

 mash

-0.64

 nicknamed

-0.61

 Mirage

-0.59

 disguised

-0.59

 Quart

-0.59

 wors

-0.58

 pyramid

-0.56

 Amon

-0.56

 mathemat

-0.56

POSITIVE LOGITS

their

0.98

them

0.95

his

0.91

him

0.88

your

0.85

those

0.82

DERR

0.81

selves

0.81

everything

0.81

sic

0.80

Activations Density 0.339%

instructions or steps related to organizing information and tasks effectively

No Comments

No Known Activations