INDEX

Explanations

numbers or quantified information

oai_token-act-pair · gpt-3.5-turbo

the word "estimates" in various contexts referring to numerical approximations or assessments

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

 Virtue

-0.74

cellence

-0.69

BOX

-0.65

usp

-0.65

itor

-0.65

 switch

-0.65

 Myst

-0.64

isode

-0.64

 privilege

-0.63

switch

-0.63

POSITIVE LOGITS

 estimates

3.71

 Estimates

2.72

 estimate

2.70

 forecasts

2.19

 estimating

2.19

 estimated

2.16

 estimation

2.06

 estim

2.05

 guesses

1.93

 projections

1.85

Activations Density 0.015%

numbers or quantified information

the word "estimates" in various contexts referring to numerical approximations or assessments

No Comments

No Known Activations