INDEX

Explanations

phrases containing the word "Tre" followed by a number

oai_token-act-pair · gpt-3.5-turbo

mentions of the name "Tre" 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 @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

 diagn

-0.59

 wastes

-0.58

INESS

-0.57

ulating

-0.57

 dated

-0.57

enegger

-0.56

 scares

-0.56

 spectrum

-0.56

 Parenthood

-0.56

ulators

-0.55

POSITIVE LOGITS

asures

1.27

asury

1.21

acher

1.19

llo

1.15

asured

1.14

ngth

1.13

von

1.11

asure

1.09

bil

1.04

asuring

0.98

Activations Density 0.029%

phrases containing the word "Tre" followed by a number

mentions of the name "Tre" in various contexts

No Comments

No Known Activations