INDEX

Explanations

percentages and rankings related to different categories

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 endif

-0.78

ĺħ

-0.75

 cannabin

-0.73

 proble

-0.72

 nomine

-0.69

 recip

-0.69

ĸļ

-0.68

endif

-0.68

 mosqu

-0.67

LET

-0.65

POSITIVE LOGITS

ãĥĲ

0.87

votes

0.83

Uncommon

0.80

Battle

0.80

neutral

0.79

ãĥ¯ãĥ³

0.77

gamer

0.77

Ultra

0.75

Resp

0.74

Spoiler

0.74

Activations Density 0.045%

percentages and rankings related to different categories

No Comments

No Known Activations