INDEX

Explanations

locations, especially cities and states

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

ï¸ı

-0.74

ICS

-0.67

llers

-0.66

ilater

-0.65

 numbering

-0.64

 monop

-0.63

 numbered

-0.63

llor

-0.63

hower

-0.61

pus

-0.60

POSITIVE LOGITS

 Francisco

1.25

 Diego

1.22

ctuary

1.12

 Antonio

1.10

San

1.08

itary

1.03

 Bernardino

0.99

Disk

0.98

ibel

0.95

 Angeles

0.95

Activations Density 0.284%

locations, especially cities and states

No Comments

No Known Activations