INDEX

Explanations

phrases related to news articles and reports

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

asse

-0.72

inished

-0.67

inho

-0.64

Â¯Â¯

-0.64

 habitable

-0.64

hetti

-0.63

BuyableInstoreAndOnline

-0.61

 Hamm

-0.61

uate

-0.60

ueless

-0.60

POSITIVE LOGITS

worthy

1.11

flash

1.09

room

1.08

reader

1.05

feed

1.05

groups

0.99

worthiness

0.98

rooms

0.95

letters

0.94

orial

0.91

Activations Density 0.445%

phrases related to news articles and reports

No Comments

No Known Activations