INDEX

Explanations

logical connectors indicating causation or conclusion

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

ISION

-0.75

igger

-0.75

WN

-0.67

quished

-0.65

iage

-0.65

umbn

-0.65

uggets

-0.64

 Colombian

-0.64

 Beaver

-0.62

UGH

-0.62

POSITIVE LOGITS

forth

1.40

 deserve

1.00

 excludes

0.95

 deserves

0.94

 qualifies

0.91

 ineligible

0.90

 qualify

0.89

 cannot

0.88

owe

0.84

 unable

0.82

Activations Density 0.018%

logical connectors indicating causation or conclusion

No Comments

No Known Activations