INDEX

Explanations

technical terms or descriptions related to software programming and technology

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

letters

-0.75

usalem

-0.69

apons

-0.66

ujah

-0.65

psey

-0.65

assad

-0.65

called

-0.64

untled

-0.64

orously

-0.63

rose

-0.63

POSITIVE LOGITS

 considering

1.11

 behold

0.90

 compared

0.79

Especially

0.76

especially

0.74

 hindsight

0.72

 Especially

0.71

huh

0.71

 judging

0.69

 nonetheless

0.69

Activations Density 3.971%

technical terms or descriptions related to software programming and technology

No Comments

No Known Activations