INDEX

Explanations

percentages and their representation in various contexts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res_post_32k-oai

Configuration

jbloom/GPT2-Small-OAI-v5-32k-resid-post-SAEs/v5_32k_layer_10.pt

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

32,768

Data Type

torch.float32

Hook Name

blocks.10.hook_resid_post

Hook Layer

Architecture

standard

Context Size

Dataset

Skylion007/openwebtext

Activation Function

topk

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 commuting

-0.68

 Rohingya

-0.66

 Patel

-0.65

 Bengal

-0.65

 knowing

-0.64

 Spit

-0.63

 Stead

-0.62

 dogged

-0.61

 Mubarak

-0.61

 Labrador

-0.60

POSITIVE LOGITS

wind

1.14

AppData

1.14

temp

1.11

loop

0.94

module

0.93

WIND

0.92

attr

0.92

prep

0.92

main

0.91

USER

0.90

Activations Density 0.143%

percentages and their representation in various contexts

No Comments

No Known Activations