INDEX

Explanations

hashtags used on social media

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

kus

-0.93

 destro

-0.75

 trave

-0.72

 contracting

-0.71

 seiz

-0.71

ierrez

-0.67

 dece

-0.65

odcast

-0.65

 conclud

-0.64

 crumbling

-0.63

POSITIVE LOGITS

 hashtag

1.35

"#

1.20

 hasht

1.18

tags

1.12

="#

0.93

âĢİ

0.84

=""

0.82

RED

0.80

"@

0.79

(#

0.78

Activations Density 0.019%

hashtags used on social media

No Comments

No Known Activations