INDEX

Explanations

words related to government leaks, whistleblowers, and organizations like WikiLeaks

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ogun

-0.78

 Solitaire

-0.75

alities

-0.74

ptive

-0.73

 Mayo

-0.72

ality

-0.69

 Krug

-0.68

uate

-0.68

relation

-0.67

 denomin

-0.66

POSITIVE LOGITS

Leaks

1.24

ileaks

1.22

 Assange

1.06

 WikiLeaks

0.98

 revelations

0.96

 trove

0.94

 Snowden

0.91

 Files

0.89

 Wikileaks

0.89

 disclosures

0.86

Activations Density 0.171%

words related to government leaks, whistleblowers, and organizations like WikiLeaks

No Comments

No Known Activations