INDEX

Explanations

abstract nouns related to specific events or concepts, like "rain," "siege," and "transition."

oai_token-act-pair · gpt-3.5-turbo

terms related to quantities, measurements, and time references

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.5.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

orks

-0.64

 Conserv

-0.62

 Skies

-0.61

...]

-0.60

 gluc

-0.60

Tea

-0.59

putable

-0.58

lihood

-0.58

given

-0.56

nu

-0.56

POSITIVE LOGITS

 laun

0.75

agall

0.70

 apiece

0.67

 starter

0.66

 commute

0.66

 bonded

0.65

itent

0.65

 sandwic

0.64

ATURES

0.63

hemoth

0.63

Activations Density 0.291%

abstract nouns related to specific events or concepts, like "rain," "siege," and "transition."

terms related to quantities, measurements, and time references

No Comments

No Known Activations