INDEX

Explanations

-word pairs with a structure like "dash" and "word"

oai_token-act-pair · gpt-3.5-turbo

references to data analysis and data-driven concepts

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

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 vetoed

-0.76

 trem

-0.76

 Hutchinson

-0.75

ampions

-0.75

 Bentley

-0.75

Wel

-0.72

 Newport

-0.72

 Revival

-0.71

 Kaine

-0.71

 Lamp

-0.71

POSITIVE LOGITS

 datasets

2.07

 dataset

2.06

 data

2.06

 Data

2.01

 DATA

1.91

Data

1.89

data

1.79

 analytics

1.77

 databases

1.61

DATA

1.58

Activations Density 0.429%

-word pairs with a structure like "dash" and "word"

references to data analysis and data-driven concepts

No Comments

No Known Activations