INDEX

Explanations

numerical rankings or positions associated with various subjects

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

ãĥ¥

-0.79

 answ

-0.67

bably

-0.65

 compan

-0.63

humane

-0.62

oat

-0.61

ãĤ©

-0.61

 rall

-0.61

 bait

-0.59

 guiActiveUn

-0.59

POSITIVE LOGITS

 Syndicate

0.75

 Committees

0.73

 Panc

0.65

 Incarn

0.64

 Frames

0.64

 Drawing

0.64

 Locked

0.63

 Finally

0.63

[+]

0.63

 Models

0.62

Activations Density 0.044%

numerical rankings or positions associated with various subjects

No Comments

No Known Activations