INDEX

Explanations

words related to personal interactions and conversations

oai_token-act-pair · gpt-3.5-turbo

references to political confrontations or conflicts involving prominent figures

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

LTD

-0.72

MAP

-0.66

 Architects

-0.66

 Advance

-0.65

 Locations

-0.63

 partnership

-0.62

thren

-0.61

 pioneers

-0.61

 Mehran

-0.61

POSITIVE LOGITS

 sarcast

1.20

 yelling

1.17

 whining

1.15

 complaining

1.10

 accusing

1.07

 inco

1.06

 awkwardly

1.06

 angrily

1.05

 mocking

1.05

 apologizing

1.05

Activations Density 0.789%

words related to personal interactions and conversations

references to political confrontations or conflicts involving prominent figures

No Comments

No Known Activations