INDEX

Explanations

proper nouns, probably related to history or politics

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

PER

-0.89

ĸļ

-0.84

pheus

-0.80

igslist

-0.79

ĺħ

-0.76

raltar

-0.74

 Pastebin

-0.72

uyomi

-0.68

phis

-0.68

PROV

-0.67

POSITIVE LOGITS

ragon

0.93

robe

0.93

ice

0.92

furt

0.85

rill

0.85

ians

0.83

stown

0.83

ard

0.83

 Mechdragon

0.81

rums

0.79

Activations Density 0.039%

proper nouns, probably related to history or politics

No Comments

No Known Activations