INDEX

Explanations

phrases related to errors or faults

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

rollers

-0.81

well

-0.73

amen

-0.73

electric

-0.73

azine

-0.70

bians

-0.69

zona

-0.69

weeney

-0.68

agonists

-0.67

ramid

-0.67

POSITIVE LOGITS

 mistakes

0.92

 mistake

0.82

 perpetrated

0.76

fulness

0.74

 careless

0.74

 mistaken

0.73

 mishand

0.73

 Gamble

0.72

é»Ĵ

0.71

 gamb

0.70

Activations Density 0.030%

phrases related to errors or faults

No Comments

No Known Activations