INDEX

Explanations

people described in a narrative, particularly those who are not identified by name

oai_token-act-pair · gpt-3.5-turbo

phrases that include the word "who" indicating anonymous sources or individuals

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

 grounding

-0.65

entry

-0.61

iscover

-0.61

luster

-0.61

ulp

-0.60

 guiActiveUnfocused

-0.59

 formation

-0.59

 plank

-0.59

tackle

-0.58

course

-0.58

POSITIVE LOGITS

 wished

1.17

 spoke

1.04

 identifies

1.03

 cannot

1.00

 declined

0.98

 prefers

0.98

 asked

0.97

 wishes

0.96

 requested

0.92

 identified

0.91

Activations Density 0.084%

people described in a narrative, particularly those who are not identified by name

phrases that include the word "who" indicating anonymous sources or individuals

No Comments

No Known Activations