INDEX

Explanations

references to tragic events, conflicts, and investigations

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

opia

-0.78

ageddon

-0.73

ocracy

-0.70

angle

-0.70

ãĤ¹ãĥĪ

-0.69

BuyableInstoreAndOnline

-0.67

 Himself

-0.65

omo

-0.65

ãĤĭ

-0.64

senal

-0.64

POSITIVE LOGITS

 observers

1.08

 commentators

1.04

 commenters

1.00

 analysts

1.00

 scholars

0.95

 researchers

0.90

 individuals

0.90

 reviewers

0.87

 people

0.86

 authors

0.85

Activations Density 1.441%

references to tragic events, conflicts, and investigations

No Comments

No Known Activations