INDEX

Explanations

numerical quantities or statistics in the form of numbers

oai_token-act-pair · gpt-3.5-turbo

phrases that quantify certain statistics or numbers

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

ĺ

-0.70

bet

-0.64

rust

-0.64

ICA

-0.64

ï¿½ï¿½ï¿½ï¿½

-0.64

HF

-0.63

IER

-0.62

 srfAttach

-0.62

iste

-0.62

hol

-0.61

POSITIVE LOGITS

 allowable

0.84

 digits

0.78

 entries

0.76

 times

0.69

 entrants

0.68

 mileage

0.67

 servings

0.66

utenberg

0.65

 occurrences

0.64

 pounds

0.64

Activations Density 0.079%

numerical quantities or statistics in the form of numbers

phrases that quantify certain statistics or numbers

No Comments

No Known Activations