INDEX

Explanations

words related to immersion or complete involvement

oai_token-act-pair · gpt-3.5-turbo

instances of the word "es," indicating a focus on verb forms or grammatical structures

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 3-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

lyak

-0.66

ylum

-0.65

rogens

-0.65

 Sinai

-0.62

oples

-0.61

romising

-0.61

 Ãľ

-0.61

hou

-0.59

 Amon

-0.59

 yeast

-0.58

POSITIVE LOGITS

Nav

0.79

conserv

0.68

Notable

0.67

Buy

0.67

Math

0.66

Vill

0.65

isSpecialOrderable

0.63

é£

0.62

ãĥ¬

0.62

Font

0.62

Activations Density 0.000%

words related to immersion or complete involvement

instances of the word "es," indicating a focus on verb forms or grammatical structures

No Comments

No Known Activations