INDEX

Explanations

words related to starting or initiating actions, such as "start," "begin," etc

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

obi

-0.70

aths

-0.69

 entirety

-0.65

itsch

-0.65

athed

-0.62

ugs

-0.61

ighth

-0.61

phy

-0.61

owl

-0.60

rero

-0.59

POSITIVE LOGITS

 anew

1.31

nings

0.94

 behaving

0.83

 raining

0.82

rek

0.75

 bothering

0.73

ribune

0.72

ners

0.71

 accumulating

0.70

 airing

0.70

Activations Density 0.450%

words related to starting or initiating actions, such as "start," "begin," etc

No Comments

No Known Activations