INDEX

Explanations

the word "to" at a high activation value

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

hop

-0.62

 folded

-0.62

typ

-0.59

hops

-0.57

 refunds

-0.54

 flakes

-0.54

 rolled

-0.53

 suspended

-0.53

 folds

-0.53

 repeats

-0.52

POSITIVE LOGITS

ggles

1.25

 ensure

1.17

ilet

1.16

 achieve

1.06

 maximize

1.06

wered

1.06

 avoid

1.05

 begin

1.02

 minimize

1.02

iling

1.01

Activations Density 0.371%

the word "to" at a high activation value

No Comments

No Known Activations