INDEX

Explanations

items or products that are being rated or discussed

oai_token-act-pair · gpt-3.5-turbo

repeated references to a specific item in a categorization or listing context

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 2-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.2.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.2.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

-0.85

inav

-0.66

osate

-0.66

 Zimmer

-0.65

 Roads

-0.61

 Curve

-0.61

 Lanka

-0.60

 Schro

-0.60

doms

-0.60

rome

-0.60

POSITIVE LOGITS

izer

0.97

aclysm

0.93

item

0.93

 item

0.92

 items

0.86

iques

0.84

potion

0.83

handler

0.81

Condition

0.81

ized

0.80

Activations Density 0.013%

items or products that are being rated or discussed

repeated references to a specific item in a categorization or listing context

No Comments

No Known Activations