INDEX

Explanations

words related to controversies or disputes

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

tackle

-0.67

mostly

-0.65

ires

-0.64

leted

-0.63

 Conan

-0.59

now

-0.58

probably

-0.58

 horm

-0.58

 Reson

-0.58

Mit

-0.57

POSITIVE LOGITS

 anything

1.23

chard

1.16

any

1.07

nam

1.07

Else

1.04

acles

1.01

 else

1.00

acle

0.98

 anymore

0.95

chid

0.94

Activations Density 0.108%

words related to controversies or disputes

No Comments

No Known Activations