INDEX

Explanations

technical terms and code-related phrases, including hyphens and numbers

oai_token-act-pair · gpt-3.5-turbo

negative sentiments or phrases that denote conflicts

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

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 10-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.10.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.10.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

theless

-0.83

 slog

-0.82

 catast

-0.80

 penal

-0.80

 tentacles

-0.80

etheless

-0.79

 deliber

-0.77

 icing

-0.77

 nursing

-0.77

 hydra

-0.77

POSITIVE LOGITS

Own

1.50

Country

1.49

Specific

1.46

Year

1.43

Face

1.42

Aid

1.41

Tech

1.41

Cent

1.41

Dri

1.40

Fit

1.40

Activations Density 0.101%

technical terms and code-related phrases, including hyphens and numbers

negative sentiments or phrases that denote conflicts

No Comments

No Known Activations