INDEX

Explanations

languages

oai_token-act-pair · gpt-3.5-turbo

mentions of the term "Cant" and its variations, indicating a focus on a specific cultural or linguistic reference

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 1-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.1.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.1.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

Reviewer

-0.79

ãģį

-0.73

Jess

-0.71

CVE

-0.71

 è£ıè

-0.67

issance

-0.66

*/(

-0.66

vernment

-0.66

IGHTS

-0.66

aphael

-0.65

POSITIVE LOGITS

 Cant

1.17

illon

1.01

ardi

0.81

ford

0.78

aret

0.77

rell

0.76

onso

0.76

asia

0.75

omp

0.72

ina

0.72

Activations Density 0.006%

languages

mentions of the term "Cant" and its variations, indicating a focus on a specific cultural or linguistic reference

No Comments

No Known Activations