INDEX

Explanations

rankings and statistical measures related to voting and participation

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

rote

-0.74

wings

-0.67

erc

-0.61

ebus

-0.61

RIPT

-0.60

 breat

-0.59

owe

-0.59

apers

-0.58

 Whip

-0.57

©¶æ¥µ

-0.56

POSITIVE LOGITS

isSpecialOrderable

0.98

 probabilities

0.84

DragonMagazine

0.80

 positives

0.79

 ascending

0.75

0.73

channelAvailability

0.73

 sidx

0.72

+(

0.71

Activations Density 0.109%

rankings and statistical measures related to voting and participation

No Comments

No Known Activations