INDEX

Explanations

numeric values part of a sequence

oai_token-act-pair · gpt-3.5-turbo

phrases related to the analysis of data or metrics

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

etheless

-0.90

âĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢâĶĢ

-0.84

albeit

-0.76

================================================================

-0.76

AFP

-0.75

âĢ

-0.75

 nonetheless

-0.75

Labour

-0.75

Â¶

-0.75

 nevertheless

-0.74

POSITIVE LOGITS

 favorites

0.65

 flyers

0.63

 Hispanic

0.61

 brainstorm

0.58

 Hispanics

0.58

 nickname

0.57

 sponsors

0.57

 pairs

0.57

 initials

0.57

 magazines

0.56

Activations Density 1.310%

numeric values part of a sequence

phrases related to the analysis of data or metrics

No Comments

No Known Activations