INDEX

Explanations

names, particularly the name "Tim"

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

 Nile

-0.71

 Angels

-0.62

holder

-0.61

 shaft

-0.61

backer

-0.60

smith

-0.59

 phosphate

-0.58

 Journals

-0.58

 asteroids

-0.58

 Wilkinson

-0.58

POSITIVE LOGITS

eless

1.57

itives

1.24

ework

1.24

elines

1.22

eworks

1.21

etime

1.17

eness

1.16

itionally

1.16

orously

1.15

elight

1.15

Activations Density 1.097%

names, particularly the name "Tim"

No Comments

No Known Activations