INDEX

Explanations

phrases related to Q&A or question and answer sessions

oai_token-act-pair · gpt-3.5-turbo

occurrences of the "Q&A" format in discussions or texts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 9-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.9.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.9.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

ufact

-0.78

raint

-0.60

etheless

-0.59

 inev

-0.57

enance

-0.57

 peach

-0.55

 poorer

-0.55

Suc

-0.54

 Hutchinson

-0.53

 phased

-0.53

POSITIVE LOGITS

FAQ

0.92

IONS

0.84

0.82

Ds

0.77

º

0.77

answered

0.77

Answer

0.76

Cs

0.75

answer

0.75

qs

0.73

Activations Density 0.020%

phrases related to Q&A or question and answer sessions

occurrences of the "Q&A" format in discussions or texts

No Comments

No Known Activations