INDEX

Explanations

phrases related to technology and product features

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

iuses

-0.70

interstitial

-0.68

 mans

-0.65

 vicious

-0.64

 pearl

-0.64

 scree

-0.59

emet

-0.59

 dehuman

-0.59

cellence

-0.58

 glor

-0.58

POSITIVE LOGITS

 Ideally

1.33

Assuming

1.24

 Alternatively

1.23

Either

1.16

 Hopefully

1.15

 Assuming

1.15

 preferably

1.14

 Otherwise

1.14

 Either

1.12

Otherwise

1.10

Activations Density 3.037%

phrases related to technology and product features

No Comments

No Known Activations