INDEX

Explanations

references to numerical values related to rankings or categorizations, often featuring a hyphen before the number

oai_token-act-pair · gpt-3.5-turbo

negative phrases related to seasons 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 @ 11-res-jb

Configuration

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

hus

-0.69

 Clockwork

-0.67

 Shogun

-0.63

 Shakespeare

-0.63

Zo

-0.63

 Solid

-0.63

 reins

-0.61

 Shib

-0.61

 Territories

-0.61

 Earthqu

-0.61

POSITIVE LOGITS

season

1.20

duty

1.19

trial

1.15

issue

1.14

sized

1.13

level

1.13

strength

1.11

edition

1.08

area

1.05

course

1.04

Activations Density 0.065%

references to numerical values related to rankings or categorizations, often featuring a hyphen before the number

negative phrases related to seasons in various contexts

No Comments

No Known Activations