INDEX

Explanations

terms related to upcoming or upcoming events

oai_token-act-pair · gpt-3.5-turbo

the word "Up" in various contexts

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

 vain

-0.76

 plain

-0.72

 judgment

-0.69

 recess

-0.68

 sarc

-0.68

 disgust

-0.68

 sanct

-0.67

 deserts

-0.67

 bore

-0.66

 medi

-0.66

POSITIVE LOGITS

Up

3.49

up

2.03

UP

1.99

Down

1.82

ups

1.63

Up

1.60

Upgrade

1.34

DOWN

1.33

Forward

1.30

UP

1.22

Activations Density 0.010%

terms related to upcoming or upcoming events

the word "Up" in various contexts

No Comments

No Known Activations