INDEX

Explanations

phrases related to locations or events

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

ĳå£«

-0.72

DragonMagazine

-0.71

 Palmer

-0.64

ooth

-0.62

ocene

-0.62

 coefficients

-0.61

ramid

-0.61

 reconc

-0.60

å£

-0.60

 fellows

-0.59

POSITIVE LOGITS

viron

1.47

igma

1.13

chanted

1.10

 masse

1.10

quire

1.08

unci

1.08

semb

1.08

cyclop

1.07

forcer

1.06

amel

1.05

Activations Density 0.009%

phrases related to locations or events

No Comments

No Known Activations