INDEX

Explanations

references to different layers in various contexts

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 8-res_fs49152-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

49,152

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

ãĥīãĥ©

-0.88

orie

-0.80

date

-0.78

ILLE

-0.78

 Predators

-0.74

 Springer

-0.73

lishing

-0.73

etheus

-0.73

photo

-0.71

itarian

-0.70

POSITIVE LOGITS

 layers

1.34

 layer

1.14

 thickness

1.11

layer

0.83

 opacity

0.81

 layered

0.80

 veil

0.80

 separ

0.79

 refinement

0.76

coat

0.75

Activations Density 0.012%

references to different layers in various contexts

No Comments

No Known Activations