INDEX

Explanations

phrases related to legal actions or discussions, like pressing charges and arguing

oai_token-act-pair · gpt-3.5-turbo

phrases relating to actions and decision-making

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

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

 Uriel

-0.68

 embodied

-0.66

arcity

-0.63

 narrower

-0.62

 intimately

-0.61

 strikingly

-0.60

 articulated

-0.58

 firsthand

-0.57

 dwar

-0.57

 Emanuel

-0.57

POSITIVE LOGITS

fuck

0.92

 haha

0.88

Fuck

0.84

 anymore

0.84

 crap

0.80

 shit

0.79

Sorry

0.79

shit

0.79

Fail

0.79

Done

0.79

Activations Density 0.464%

phrases related to legal actions or discussions, like pressing charges and arguing

phrases relating to actions and decision-making

No Comments

No Known Activations