INDEX

Explanations

names related to various characters or entities, possibly from literature or folklore

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

 Colleges

-0.71

 divided

-0.69

 stadiums

-0.69

 dungeons

-0.69

 theaters

-0.68

 downgrade

-0.67

 airports

-0.67

CONCLUS

-0.67

 capitals

-0.65

 checkout

-0.65

POSITIVE LOGITS

berto

1.12

ilyn

1.11

frey

1.10

ergus

1.03

imir

1.02

isha

1.02

ileen

1.00

abeth

1.00

jan

0.99

manuel

0.99

Activations Density 1.359%

names related to various characters or entities, possibly from literature or folklore

No Comments

No Known Activations