INDEX

Explanations

news agency names at the end of articles

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Exhibition

-0.63

irlf

-0.62

req

-0.60

 spoiler

-0.56

 Sims

-0.56

 undergrad

-0.55

 Stall

-0.55

ozo

-0.54

bage

-0.54

course

-0.52

POSITIVE LOGITS

eka

0.77

Reuters

0.68

 WORLD

0.67

 sidx

0.65

 Kurdistan

0.64

heast

0.63

vana

0.62

everal

0.62

ANCE

0.61

 WATCHED

0.60

Activations Density 0.023%

news agency names at the end of articles

No Comments

No Known Activations