INDEX

Explanations

declarations of error or incorrectness in rationale or beliefs

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

apest

-0.72

 concess

-0.71

egu

-0.67

estern

-0.59

yss

-0.58

Mub

-0.58

psey

-0.56

ileged

-0.56

orkshire

-0.54

earchers

-0.54

POSITIVE LOGITS

;)

0.75

 imaginable

0.73

 attRot

0.69

 oneself

0.67

*.

0.66

?).

0.66

 haha

0.65

 itself

0.64

 existed

0.64

 herself

0.63

Activations Density 0.381%

declarations of error or incorrectness in rationale or beliefs

No Comments

No Known Activations