I have a local LLM, GPT4ALL with a Mistral OpenOrca model (a...

原始 JSON

{
  "kind": 1,
  "id": "40001016c0c399ae86236e1e6510a3cbc40d9280c4b2094fb504f1f5b9589232",
  "pubkey": "9a21569255d0a3a9e75f1de2e4c883c9be2e5615887f22b2ecf6b1813bcd587d",
  "created_at": 1764795373,
  "tags": [
    [
      "p",
      "55f573b651eff351db57b0601d23022d8c532f9825db10a5733ebf39be4aa21b",
      "wss://relay.sovereign.app/",
      "Comte de Sats Germain"
    ],
    [
      "e",
      "9be783927c98c2bef9d703fc6a06dbbe78a19f31a2a3bde7404774f48d4ff58d",
      "wss://relay.sovereign.app/",
      "root",
      "55f573b651eff351db57b0601d23022d8c532f9825db10a5733ebf39be4aa21b"
    ]
  ],
  "content": "I have a local LLM, GPT4ALL with a Mistral OpenOrca model (about all my embedded AMD GPU can handle on my GNU/Linux box). I'll put some of my cross-domain \u003c 300 char posts into it to see what the matching returns. 99.99% of the time the posts get zero engagement on any channel (Bluesky, Mastodon, Nostr), so I just want to see if my own correlation engine is wackadoodle. The LLM usually picks up on my cross-domain correlations and reflects them back accurately.  Usually in the .001% of the time a human *does* respond to my cross-domain ~ 300 character posts, they are still doing a pattern match and bringing in their own suitcase of experience and prejudices. This is a very human behavior that I do myself.  Vetting it against the generic model is useful.   Plus, since it is GPT4ALL, it is completely local (unless I publish it in my journal or on some open Nostr relays, haha).\n\nI also like it that it saves my prompts. Related to our brief back and forth today on myth, late last spring I entered this prompt:\n\n\"The idea of Daniel Quinn's Ishmael, that you can identify myth in culture that nobody in your culture sees as myth, and that somehow this will help you, particularly in a culture with \u003c 3 relation cognition, exposes readers to a full-on Gregory Bateson style Double Bind. Schizophrenia\"\n\nThis returned:\n\n\"Ishmael is a novel by Daniel Quinn that presents the idea of identifying myths in culture, which are often unrecognized as such. The protagonist, an ape named Ishmael, teaches his human student about the history and beliefs of various cultures throughout time.\n\nThe concept of \"less than 3 relation cognition\" refers to a person's ability to understand relationships between different elements or concepts in their environment. In this context, it seems to suggest that people may have difficulty recognizing myths within their own culture because they lack the cognitive capacity to perceive these connections and patterns.\n\nThe idea of exposing readers to a \"Double Bind\" and Schizophrenia is not explicitly mentioned in Ishmael but can be related to the novel's theme of challenging one's beliefs and understanding of reality. A double bind occurs when an individual receives conflicting messages or expectations, leading them into a paradoxical situation with no clear solution. In this case, readers may feel confused or overwhelmed as they attempt to reconcile their own cultural myths and beliefs with the alternative perspectives presented by Ishmael.\n\nSchizophrenia is a mental disorder characterized by distorted thoughts, perceptions, emotions, and behaviors. While it is not directly related to the concept of identifying myth in culture, the idea of exposing readers to conflicting messages or challenging their understanding of reality may evoke feelings similar to those experienced by individuals with schizophrenia. This can be seen as a metaphorical way of describing the cognitive dissonance that arises when confronted with new and potentially disruptive information about one's own culture and beliefs.\"\n\n\nAt a minimum, it gives me clues as to where the confusion lays, and I can make my connections clearer.  One thing about humans and shared cultural cognition, is as awesome as we are at it, we are still limited to roughly two teams and three tools... all of us [1]. Maps and rulesets change this.  You and I have a slightly different definition of maps and rulesets, I seem to remember. I separate maps and rulesets from the underlying processes and structure. I think of it like a literal map. Without one you tend to go in circles in the woods [2].\n\nBut, yeah, I can see how this also appeals to ego, as it is essentially a validator.  \n\n[1] https://web.archive.org/web/20070418142809/http://groups.inf.ed.ac.uk/ppig/readinglist/tomasello_et_al2004.pdf\n\n[2] https://doi.org/10.1016/j.cub.2009.07.053",
  "sig": "148e679e5b0ae0a4f6fba7e9e13c4be445ec12bf5af966bb99c66f6c2bb07dfc00c78473f8320ba7ce35360af84c73b744d8d6596c4b2c7062b26d4a243694dd"
}