INDEX

Explanations

instances of failure or wrongdoing

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

ãĥ¼ãĥĨãĤ£

-0.72

 Variant

-0.66

BUR

-0.65

ORGE

-0.65

zer

-0.64

soType

-0.62

Strange

-0.62

tesque

-0.62

çīĪ

-0.60

 Mayhem

-0.59

POSITIVE LOGITS

 adequately

1.31

 properly

1.12

 comply

1.00

 timely

0.95

hin

0.95

 sufficiently

0.94

 assimil

0.94

 heed

0.92

 adequate

0.92

 cooperate

0.89

Activations Density 1.348%

instances of failure or wrongdoing

No Comments

No Known Activations