INDEX

Explanations

names of individuals or 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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

izoph

-0.70

ccording

-0.67

Xi

-0.62

 unfavorable

-0.60

 deadliest

-0.59

Scotland

-0.58

 polarization

-0.57

 dismal

-0.57

 manpower

-0.57

Load

-0.57

POSITIVE LOGITS

Jr

1.63

III

1.30

Sr

1.24

QC

1.11

aka

1.10

berger

1.09

baum

1.07

(@

1.05

burgh

1.03

meyer

1.01

Activations Density 2.551%

names of individuals or entities

No Comments

No Known Activations