INDEX

Explanations

imperative statements indicating a requirement or necessity

oai_token-act-pair · gpt-3.5-turbo

Triggered by @bot

modal verbs related to obligations and permissions

oai_token-act-pair · gpt-4o-mini

Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 12-res-jb

Configuration

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

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_post

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Negative Logits

 hating

-0.71

 dreaming

-0.71

 digging

-0.67

 ripping

-0.67

 trillions

-0.63

 hungry

-0.61

 frantic

-0.61

 pulling

-0.60

 tearing

-0.60

telling

-0.60

POSITIVE LOGITS

 apply

1.08

 occur

1.04

 accompany

1.02

 expire

0.98

 Apply

0.96

be

0.92

 vary

0.91

 include

0.90

 denote

0.89

 incur

0.89

Activations Density 0.165%

imperative statements indicating a requirement or necessity

modal verbs related to obligations and permissions

No Known Activations