INDEX

Explanations

terms related to unraveling or deciphering information or situations

oai_token-act-pair · gpt-3.5-turbo

processes related to understanding and breaking down complex systems

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

 reserved

-0.67

jet

-0.62

fleet

-0.61

ggle

-0.61

 planners

-0.60

eworld

-0.60

 hire

-0.60

lee

-0.60

obar

-0.60

oline

-0.59

POSITIVE LOGITS

ynski

0.88

 unravel

0.79

taining

0.78

stakes

0.77

ãĤ©

0.76

ĸļ

0.75

İĭ

0.74

schild

0.74

lations

0.74

edIn

0.72

Activations Density 0.055%

terms related to unraveling or deciphering information or situations

processes related to understanding and breaking down complex systems

No Comments

No Known Activations