INDEX

Explanations

content related to guides or step-by-step processes on keeping resolutions or making changes

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

©¶æ¥µ

-0.93

adra

-0.77

elected

-0.74

Ń·

-0.72

arak

-0.70

nesty

-0.70

geoning

-0.69

wen

-0.69

inction

-0.68

ittee

-0.68

POSITIVE LOGITS

 firsthand

1.12

 topics

1.10

 specifics

1.08

 similarities

1.07

 misconceptions

1.06

 pros

1.02

 workings

1.02

 details

1.02

 insights

1.02

 perspectives

1.01

Activations Density 7.233%

content related to guides or step-by-step processes on keeping resolutions or making changes

No Comments

No Known Activations