INDEX

Explanations

specific locations and entities

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

bilt

-0.81

andise

-0.74

izabeth

-0.67

 biscuits

-0.66

ndum

-0.65

 Seym

-0.65

¬¼

-0.63

 Guinness

-0.63

 silence

-0.62

£ı

-0.62

POSITIVE LOGITS

irection

1.27

ividual

1.12

eways

1.07

iots

1.05

ocument

1.02

imensional

1.00

gets

0.98

irect

0.96

iosyncr

0.94

der

0.93

Activations Density 0.083%

specific locations and entities

No Comments

No Known Activations