INDEX

Explanations

instances where the word 'to' is doubled

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

ISA

-0.71

rim

-0.65

ellectual

-0.61

estic

-0.60

 rall

-0.60

OTE

-0.59

 Cosponsors

-0.59

earable

-0.59

rod

-0.59

aila

-0.57

POSITIVE LOGITS

0.79

0.76

0.71

ruct

0.67

 onwards

0.65

0.64

ruary

0.64

Activations Density 1.486%

instances where the word 'to' is doubled

No Comments

No Known Activations