INDEX

Explanations

references to setting or breaking records

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_fs3072-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_3072

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

3,072

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

adin

-0.74

utterstock

-0.72

 cart

-0.69

creen

-0.68

nery

-0.67

ĳå£«

-0.66

eca

-0.65

hiro

-0.65

xit

-0.64

adh

-0.63

POSITIVE LOGITS

 breaker

1.09

keeping

1.05

breaking

1.03

 holder

0.95

 Breaker

0.91

 record

0.91

record

0.87

 Record

0.81

 Keeper

0.77

 linkage

0.77

Activations Density 0.522%

references to setting or breaking records

No Comments

No Known Activations