INDEX

Explanations

references to physical interaction or control

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

ļéĨĴ

-0.67

vation

-0.64

 subp

-0.63

-0.62

 Atmosp

-0.62

ITAL

-0.62

 Scotia

-0.59

Marginal

-0.58

çĶ°

-0.58

 è£ıç

-0.58

POSITIVE LOGITS

pring

1.12

maid

0.97

chool

0.93

ets

0.93

creen

0.92

handed

0.90

bag

0.89

picked

0.87

ingers

0.84

sembly

0.81

Activations Density 0.053%

references to physical interaction or control

No Comments

No Known Activations