INDEX

Explanations

comparisons of how different individuals treat others, focusing on terms like "comrades" and "servants."

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

New Auto-Interp

Configuration

jbloom/Gemma-2b-IT-Residual-Stream-SAEs/gemma_2b_it_blocks.12.hook_resid_post_16384

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

HuggingFaceFW/fineweb

Features

16,384

Data Type

float32

Hook Name

blocks.12.hook_resid_post

Hook Layer

Architecture

standard

Context Size

1,024

Dataset

Skylion007/openwebtext

Activation Function

relu

Embeds

PlotsExplanationShow Test FieldDefault Test Text

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Minang

-0.66

 disagre

-0.66

 tolerably

-0.62

 nobly

-0.62

 vainly

-0.61

 impra

-0.61

 Putih

-0.61

 profuse

-0.61

 imperfectly

-0.61

 unspeak

-0.60

POSITIVE LOGITS

 prostitu

0.67

 divertimento

0.66

 tutt

0.64

 religione

0.63

 teolog

0.62

 meras

0.59

 palio

0.59

 rilass

0.58

 fatte

0.58

 parteci

0.58

Activations Density 0.374%

comparisons of how different individuals treat others, focusing on terms like "comrades" and "servants."

No Comments

No Known Activations