INDEX

Explanations

technical and procedural language related to programming or regulations

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_fs1536-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_1536

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

1,536

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

SPONSORED

-0.69

ETF

-0.65

bian

-0.65

inea

-0.64

BSD

-0.63

igrate

-0.63

ppa

-0.62

heimer

-0.61

ivia

-0.61

actly

-0.60

POSITIVE LOGITS

 virtue

1.51

products

1.31

laws

1.13

 default

1.09

product

1.03

akuya

1.03

catch

0.98

 leaps

0.96

 mistake

0.93

gone

0.93

Activations Density 1.138%

technical and procedural language related to programming or regulations

No Comments

No Known Activations