INDEX

Explanations

phrases related to conversations or exchanges

oai_token-act-pair · gpt-3.5-turbo

conversational cues and expressions of need for communication

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 11-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.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

 WATCHED

-0.81

"#

-0.81

ĺħ

-0.74

utterstock

-0.72

Article

-0.72

xtap

-0.70

NAS

-0.69

 Zucker

-0.69

NFL

-0.68

 ï¿½ï¿½ï¿½ï¿½

-0.67

POSITIVE LOGITS

-"

2.00

âĢ¦"

1.68

..."

1.64

—"

1.59

!?"

1.35

?"

1.29

?!"

1.26

âĢ¦."

1.24

 âĢ¦"

1.22

!"

1.13

Activations Density 0.351%

phrases related to conversations or exchanges

conversational cues and expressions of need for communication

No Comments

No Known Activations