INDEX

Explanations

phrases related to law enforcement and legal proceedings

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

itiz

-0.75

 Reborn

-0.70

çĶŁ

-0.69

 Citiz

-0.67

accompan

-0.66

 Laun

-0.64

 Presence

-0.63

 Mayhem

-0.63

ersed

-0.62

 Communities

-0.61

POSITIVE LOGITS

 necessarily

1.39

 hesitate

1.23

 bother

1.22

 know

1.19

 deserve

1.17

 intend

1.14

 seem

1.14

 belong

1.11

 realise

1.09

 condone

1.07

Activations Density 0.575%

phrases related to law enforcement and legal proceedings

No Comments

No Known Activations