INDEX

Explanations

names and titles associated with prominent figures and entities

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

 prest

-0.61

 denomin

-0.52

 sugg

-0.51

 transpired

-0.50

 dwar

-0.49

 childbirth

-0.47

 warr

-0.45

 reel

-0.45

 transplant

-0.45

 reper

-0.45

POSITIVE LOGITS

 squarely

0.61

 onto

0.61

's

0.60

 differently

0.59

onto

0.58

 separately

0.56

 hostage

0.53

quin

0.52

 praises

0.52

 instead

0.51

Activations Density 0.535%

names and titles associated with prominent figures and entities

No Comments

No Known Activations