INDEX

Explanations

locations

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_fs24576-jb

Configuration

jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

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

 Rossi

-0.73

Effective

-0.62

 Consent

-0.62

Wan

-0.61

 Mehran

-0.58

 Ples

-0.57

FU

-0.57

Dee

-0.56

primary

-0.56

RECT

-0.55

POSITIVE LOGITS

emouth

1.66

oir

1.20

ourn

1.09

ette

1.06

ettes

1.04

auts

0.97

nette

0.97

esses

0.96

oise

0.94

ments

0.93

Activations Density 0.012%

locations

No Comments

No Known Activations