INDEX

Explanations

conjunctions used to connect ideas or phrases in a sentence

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

SourceFile

-0.80

ULAR

-0.62

cott

-0.61

hari

-0.60

bound

-0.60

CLOSE

-0.58

tnc

-0.57

ories

-0.57

ophone

-0.56

mates

-0.54

POSITIVE LOGITS

 alas

0.80

 beware

0.73

 nonetheless

0.72

 anecd

0.71

 unlike

0.70

 concedes

0.70

 ignores

0.69

 concede

0.69

âķĲâķĲ

0.66

 rhet

0.66

Activations Density 0.046%

conjunctions used to connect ideas or phrases in a sentence

No Comments

No Known Activations