INDEX

Explanations

mentions of control or lack of control (e.g., having power over, providing choices, setting free, imposing laws, controlling circulation)

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

hon

-0.71

ich

-0.71

eds

-0.68

à¤

-0.67

istle

-0.66

ublic

-0.65

ophers

-0.65

 à¤

-0.64

spot

-0.64

icion

-0.63

POSITIVE LOGITS

 levers

0.82

 airflow

0.80

 traffic

0.78

 impulses

0.78

ulate

0.77

ively

0.75

 pests

0.73

 tempo

0.72

 everything

0.72

 airspace

0.71

Activations Density 0.068%

mentions of control or lack of control (e.g., having power over, providing choices, setting free, imposing laws, controlling circulation)

No Comments

No Known Activations