INDEX

Explanations

words related to actions or intentions

oai_token-act-pair · gpt-3.5-turbo

instances of potential actions or capabilities expressed in the conditional or ability form

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 5-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.5.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.5.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 irony

-0.65

 understatement

-0.64

 downside

-0.63

rift

-0.60

 reperc

-0.60

RM

-0.59

 dimensional

-0.59

ARK

-0.58

 supra

-0.58

 narciss

-0.57

POSITIVE LOGITS

ilater

0.84

sylv

0.80

 safely

0.77

 satisfy

0.71

eming

0.68

 begin

0.67

shed

0.67

 stay

0.66

 continue

0.66

anqu

0.65

Activations Density 0.127%

words related to actions or intentions

instances of potential actions or capabilities expressed in the conditional or ability form

No Comments

No Known Activations