INDEX

Explanations

phrases related to investigation or thorough examination

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_fs12288-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

12,288

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

xus

-0.69

 Minotaur

-0.67

nect

-0.66

acters

-0.64

 willpower

-0.64

 Palestin

-0.61

 Founding

-0.60

 bonding

-0.60

cil

-0.59

 bonded

-0.58

POSITIVE LOGITS

stakes

1.95

restling

0.99

hurst

0.90

 away

0.84

 aside

0.81

shots

0.81

sheets

0.80

atcher

0.79

abies

0.78

atches

0.76

Activations Density 0.059%

phrases related to investigation or thorough examination

No Comments

No Known Activations