INDEX

Explanations

phrases related to legal or political controversy

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

 troubles

-0.70

 delinquent

-0.65

 overl

-0.65

 trouble

-0.65

 controversy

-0.63

 swaps

-0.62

 contiguous

-0.61

 stabilized

-0.60

 giveaway

-0.59

 popularity

-0.59

POSITIVE LOGITS

 "âĢ¦

1.33

"[

1.26

 "...

1.25

"'

1.14

'[

1.10

""

1.05

"@

1.04

"(

1.01

".

0.98

"#

0.96

Activations Density 0.090%

phrases related to legal or political controversy

No Comments

No Known Activations