Gemini Experimental 1114

[shankarvashist]

Can you please add new model Gemini Experimental 1114, which has been released recently. Thanks.

[jochenkirstaetter]

Hello @shankarvashist,

Thanks for the hint, yes, will update it today.

FYI, you can also pass the model name as a string into the methods. The static class Model with the constants is a trick of laziness only. 😉

[jochenkirstaetter]

Hi @shankarvashist

Also, I'm a bit intrigued by the following information when you query the details of the model.

      "name": "models/gemini-exp-1114",
      "version": "exp-1114",
      "displayName": "Gemini Experimental 1114",
      "description": "Mid-size multimodal model that supports up to 2 million tokens",
      "inputTokenLimit": 32767,

Although the description speaks of 2M tokens, the input limit seems to be 32k only.

[jochenkirstaetter]

Hi @shankarvashist

New release v1.9.2 has been released containing the new experimental model as constant.

Thanks, JoKi

[shankarvashist]

Hi Jochen, Thanks for your reply. Well noted that the string can be passed :-) Also, I see in your other response that 1.9.2 has the new model added. This model, in google ai studio clearly specifies 32K tokens only. Seems they are experimenting with another approach which may be more resource intensive.. I'm still checking this one out.. Many Thanks.. Best Regards, Shankar Vashist.

…

On Mon, Nov 18, 2024 at 4:12 PM Jochen Kirstätter ***@***.***> wrote: Hello @shankarvashist <https://github.com/shankarvashist>, Thanks for the hint, yes, will update it today. FYI, you can also pass the model name as a string into the methods. The static class Model with the constants is a trick of laziness only. 😉 — Reply to this email directly, view it on GitHub <#45 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQAM7XW6AU3FH3A5WFK3I5L2BGOPBAVCNFSM6AAAAABR3NXUEWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIOBSGIZDMOJYGY> . You are receiving this because you were mentioned.Message ID: ***@***.***>

[jochenkirstaetter]

Hi @shankarvashist

This new experimental model also has only 32k input tokens according to the model details.

      "name": "models/gemini-exp-1114",
      "version": "exp-1114",
      "displayName": "Gemini Experimental 1114",
      "description": "Mid-size multimodal model that supports up to 2 million tokens",
      "inputTokenLimit": 32767,
      "outputTokenLimit": 8192,
      "supportedGenerationMethods": [
        "generateContent",
        "countTokens"
      ],
      "temperature": 1,
      "topP": 0.95,
      "topK": 64,
      "maxTemperature": 2

However, the description is wrong. It says 2 million tokens... 😃