INDEX

Explanations

words related to individuals or locations mentioned in news articles

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

âĸ¬âĸ¬

-0.65

zona

-0.65

 cradle

-0.64

ãĥīãĥ©ãĤ´ãĥ³

-0.63

catentry

-0.61

\\\\\\\\

-0.61

 Skydragon

-0.60

 calories

-0.59

 Sabha

-0.59

 Pradesh

-0.59

POSITIVE LOGITS

enhagen

1.09

wald

1.03

schild

1.02

enberg

1.00

enegger

0.99

lein

0.97

velt

0.95

enthal

0.92

itsch

0.91

elin

0.90

Activations Density 0.090%

words related to individuals or locations mentioned in news articles

No Comments

No Known Activations