INDEX

Explanations

words related to creating, building, and developing entities, systems, or relationships

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_fs768-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

768

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

osate

-0.78

played

-0.66

pling

-0.66

channelAvailability

-0.63

hibition

-0.63

beck

-0.63

pha

-0.62

 Unsure

-0.61

most

-0.61

andra

-0.60

POSITIVE LOGITS

 prototypes

1.12

 anew

0.93

smanship

0.92

 scaff

0.84

 customized

0.84

 prototype

0.83

 makeshift

0.82

 elaborate

0.81

urally

0.80

new

0.79

Activations Density 4.794%

words related to creating, building, and developing entities, systems, or relationships

No Comments

No Known Activations