INDEX

Explanations

phrases related to alignment or positioning

oai_token-act-pair · gpt-3.5-turbo

phrases that indicate "lines of" or "line of" 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

jee

-0.71

gdala

-0.67

»Ĵ

-0.66

paces

-0.65

é¾

-0.63

 Publishers

-0.63

 sqor

-0.62

huge

-0.61

rices

-0.61

ful

-0.61

POSITIVE LOGITS

 scrimmage

1.19

 succession

0.96

 sight

0.91

 questioning

0.82

 descent

0.79

 reasoning

0.75

 Sight

0.70

 attack

0.69

 inquiry

0.69

iasis

0.65

Activations Density 0.047%

phrases related to alignment or positioning

phrases that indicate "lines of" or "line of" in various contexts

No Comments

No Known Activations