INDEX

Explanations

occurrences of the word "reply" in different contexts

oai_token-act-pair · gpt-3.5-turbo

mentions of replies to comments or posts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 0-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.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

stakes

-0.75

elson

-0.75

veh

-0.69

Ĥª

-0.67

licts

-0.65

 impacted

-0.65

 eroded

-0.62

arc

-0.61

 illegally

-0.61

ewater

-0.61

POSITIVE LOGITS

 reply

3.91

 replies

2.71

reply

2.59

Reply

2.07

 response

2.01

 answer

1.97

 Reply

1.95

 replied

1.85

 respond

1.69

Answer

1.68

Activations Density 0.012%

occurrences of the word "reply" in different contexts

mentions of replies to comments or posts

No Comments

No Known Activations