INDEX

Explanations

references to achievements or evaluations of performances

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

ships

-0.70

agan

-0.69

ship

-0.69

 Wonderland

-0.66

lly

-0.65

pora

-0.62

 settlers

-0.61

 detachment

-0.59

UAL

-0.58

OVA

-0.58

POSITIVE LOGITS

card

1.15

cards

1.15

ific

1.07

heet

0.84

ificant

0.82

rs

0.82

emi

0.81

ifications

0.78

icket

0.76

ificate

0.74

Activations Density 0.843%

references to achievements or evaluations of performances

No Comments

No Known Activations