INDEX

Explanations

statements where the subject is described or linked to an evaluation

oai_token-act-pair · gpt-3.5-turbo Triggered by @bot

statements indicating the significance or importance of a subject

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

Not in Any Lists

No Comments

Negative Logits

ngth

-0.90

luaj

-0.80

estial

-0.72

azine

-0.70

ogether

-0.70

ventures

-0.69

aneers

-0.69

enne

-0.68

endar

-0.68

heid

-0.68

POSITIVE LOGITS

 evidenced

1.14

 because

1.12

why

1.08

 partly

1.06

 especially

1.05

 compounded

1.05

 reflected

1.04

 exacerbated

1.04

 despite

1.03

 understandable

1.02

Activations Density 0.152%

statements where the subject is described or linked to an evaluation

statements indicating the significance or importance of a subject

No Comments

No Known Activations