INDEX

Explanations

mentions of large quantities of entities or activities

oai_token-act-pair · gpt-3.5-turbo

instances of the word "scores" followed by numerical values or references to quantities

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

Forge

-0.67

ned

-0.65

 deduction

-0.63

ulkan

-0.62

 dissolution

-0.62

UAL

-0.62

iator

-0.61

Correction

-0.60

 necessity

-0.59

does

-0.59

POSITIVE LOGITS

paces

0.99

dozen

0.93

poons

0.93

imilar

0.90

 thousand

0.88

 dozen

0.83

omething

0.82

arnaev

0.80

everal

0.78

chool

0.78

Activations Density 0.016%

mentions of large quantities of entities or activities

instances of the word "scores" followed by numerical values or references to quantities

No Comments

No Known Activations