You can not select more than 25 topics Topics must start with a chinese character,a letter or number, can include dashes ('-') and can be up to 35 characters long.

ModelOptions.cs 3.1 kB

April 2024 Binary Update (#662) * Updated binaries, using [this build](https://github.com/SciSharp/LLamaSharp/actions/runs/8654672719/job/23733195669) for llama.cpp commit `f7001ccc5aa359fcf41bba19d1c99c3d25c9bcc7`. - Added all new functions. - Moved some functions (e.g. `SafeLlamaModelHandle` specific functions) into `SafeLlamaModelHandle.cs` - Exposed tokens on `SafeLlamaModelHandle` and `LLamaWeights` through a `Tokens` property. As new special tokens are added in the future they can be added here. - Changed all token properties to return nullable tokens, to handle some models not having some tokens. - Fixed `DefaultSamplingPipeline` to handle no newline token in some models. * Moved native methods to more specific locations. - Context specific things have been moved into `SafeLLamaContextHandle.cs` and made private - they're exposed through C# properties and methods already. - Checking that GPU layer count is zero if GPU offload is not supported. - Moved methods for creating default structs (`llama_model_quantize_default_params` and `llama_context_default_params`) into relevant structs. * Removed exception if `GpuLayerCount > 0` when GPU is not supported. * - Added low level wrapper methods for new per-sequence state load/save in `SafeLLamaContextHandle` - Added high level wrapper methods (save/load with `State` object or memory mapped file) in `LLamaContext` - Moved native methods for per-sequence state load/save into `SafeLLamaContextHandle` * Added update and defrag methods for KV cache in `SafeLLamaContextHandle` * Updated submodule to `f7001ccc5aa359fcf41bba19d1c99c3d25c9bcc7` * Passing the sequence ID when saving a single sequence state
1 year ago
April 2024 Binary Update (#662) * Updated binaries, using [this build](https://github.com/SciSharp/LLamaSharp/actions/runs/8654672719/job/23733195669) for llama.cpp commit `f7001ccc5aa359fcf41bba19d1c99c3d25c9bcc7`. - Added all new functions. - Moved some functions (e.g. `SafeLlamaModelHandle` specific functions) into `SafeLlamaModelHandle.cs` - Exposed tokens on `SafeLlamaModelHandle` and `LLamaWeights` through a `Tokens` property. As new special tokens are added in the future they can be added here. - Changed all token properties to return nullable tokens, to handle some models not having some tokens. - Fixed `DefaultSamplingPipeline` to handle no newline token in some models. * Moved native methods to more specific locations. - Context specific things have been moved into `SafeLLamaContextHandle.cs` and made private - they're exposed through C# properties and methods already. - Checking that GPU layer count is zero if GPU offload is not supported. - Moved methods for creating default structs (`llama_model_quantize_default_params` and `llama_context_default_params`) into relevant structs. * Removed exception if `GpuLayerCount > 0` when GPU is not supported. * - Added low level wrapper methods for new per-sequence state load/save in `SafeLLamaContextHandle` - Added high level wrapper methods (save/load with `State` object or memory mapped file) in `LLamaContext` - Moved native methods for per-sequence state load/save into `SafeLLamaContextHandle` * Added update and defrag methods for KV cache in `SafeLLamaContextHandle` * Updated submodule to `f7001ccc5aa359fcf41bba19d1c99c3d25c9bcc7` * Passing the sequence ID when saving a single sequence state
1 year ago
April 2024 Binary Update (#662) * Updated binaries, using [this build](https://github.com/SciSharp/LLamaSharp/actions/runs/8654672719/job/23733195669) for llama.cpp commit `f7001ccc5aa359fcf41bba19d1c99c3d25c9bcc7`. - Added all new functions. - Moved some functions (e.g. `SafeLlamaModelHandle` specific functions) into `SafeLlamaModelHandle.cs` - Exposed tokens on `SafeLlamaModelHandle` and `LLamaWeights` through a `Tokens` property. As new special tokens are added in the future they can be added here. - Changed all token properties to return nullable tokens, to handle some models not having some tokens. - Fixed `DefaultSamplingPipeline` to handle no newline token in some models. * Moved native methods to more specific locations. - Context specific things have been moved into `SafeLLamaContextHandle.cs` and made private - they're exposed through C# properties and methods already. - Checking that GPU layer count is zero if GPU offload is not supported. - Moved methods for creating default structs (`llama_model_quantize_default_params` and `llama_context_default_params`) into relevant structs. * Removed exception if `GpuLayerCount > 0` when GPU is not supported. * - Added low level wrapper methods for new per-sequence state load/save in `SafeLLamaContextHandle` - Added high level wrapper methods (save/load with `State` object or memory mapped file) in `LLamaContext` - Moved native methods for per-sequence state load/save into `SafeLLamaContextHandle` * Added update and defrag methods for KV cache in `SafeLLamaContextHandle` * Updated submodule to `f7001ccc5aa359fcf41bba19d1c99c3d25c9bcc7` * Passing the sequence ID when saving a single sequence state
1 year ago
April 2024 Binary Update (#662) * Updated binaries, using [this build](https://github.com/SciSharp/LLamaSharp/actions/runs/8654672719/job/23733195669) for llama.cpp commit `f7001ccc5aa359fcf41bba19d1c99c3d25c9bcc7`. - Added all new functions. - Moved some functions (e.g. `SafeLlamaModelHandle` specific functions) into `SafeLlamaModelHandle.cs` - Exposed tokens on `SafeLlamaModelHandle` and `LLamaWeights` through a `Tokens` property. As new special tokens are added in the future they can be added here. - Changed all token properties to return nullable tokens, to handle some models not having some tokens. - Fixed `DefaultSamplingPipeline` to handle no newline token in some models. * Moved native methods to more specific locations. - Context specific things have been moved into `SafeLLamaContextHandle.cs` and made private - they're exposed through C# properties and methods already. - Checking that GPU layer count is zero if GPU offload is not supported. - Moved methods for creating default structs (`llama_model_quantize_default_params` and `llama_context_default_params`) into relevant structs. * Removed exception if `GpuLayerCount > 0` when GPU is not supported. * - Added low level wrapper methods for new per-sequence state load/save in `SafeLLamaContextHandle` - Added high level wrapper methods (save/load with `State` object or memory mapped file) in `LLamaContext` - Moved native methods for per-sequence state load/save into `SafeLLamaContextHandle` * Added update and defrag methods for KV cache in `SafeLLamaContextHandle` * Updated submodule to `f7001ccc5aa359fcf41bba19d1c99c3d25c9bcc7` * Passing the sequence ID when saving a single sequence state
1 year ago
123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117
  1. using System.Text;
  2. using LLama.Abstractions;
  3. using LLama.Native;
  4. namespace LLama.Web.Common
  5. {
  6. public class ModelOptions
  7. : ILLamaParams
  8. {
  9. /// <summary>
  10. /// Model friendly name
  11. /// </summary>
  12. public string Name { get; set; }
  13. /// <summary>
  14. /// Max context insta=nces allowed per model
  15. /// </summary>
  16. public int MaxInstances { get; set; }
  17. /// <inheritdoc />
  18. public uint? ContextSize { get; set; }
  19. /// <inheritdoc />
  20. public int MainGpu { get; set; } = 0;
  21. /// <inheritdoc />
  22. public GPUSplitMode SplitMode { get; set; } = GPUSplitMode.None;
  23. /// <inheritdoc />
  24. public int GpuLayerCount { get; set; } = 20;
  25. public uint SeqMax { get; }
  26. /// <inheritdoc />
  27. public uint Seed { get; set; } = 1686349486;
  28. public bool Embeddings { get; }
  29. /// <inheritdoc />
  30. public bool UseMemorymap { get; set; } = true;
  31. /// <inheritdoc />
  32. public bool UseMemoryLock { get; set; } = false;
  33. /// <inheritdoc />
  34. public string ModelPath { get; set; }
  35. /// <inheritdoc />
  36. public AdapterCollection LoraAdapters { get; set; } = new();
  37. /// <inheritdoc />
  38. public string LoraBase { get; set; } = string.Empty;
  39. /// <inheritdoc />
  40. public uint? Threads { get; set; }
  41. /// <inheritdoc />
  42. public uint? BatchThreads { get; set; }
  43. /// <inheritdoc />
  44. public uint BatchSize { get; set; } = 512;
  45. /// <inheritdoc />
  46. public uint UBatchSize { get; set; } = 512;
  47. /// <inheritdoc />
  48. public TensorSplitsCollection TensorSplits { get; set; } = new();
  49. /// <inheritdoc />
  50. public List<MetadataOverride> MetadataOverrides { get; } = new();
  51. /// <inheritdoc />
  52. public float? RopeFrequencyBase { get; set; }
  53. /// <inheritdoc />
  54. public float? RopeFrequencyScale { get; set; }
  55. /// <inheritdoc />
  56. public float? YarnExtrapolationFactor { get; set; }
  57. /// <inheritdoc />
  58. public float? YarnAttentionFactor { get; set; }
  59. /// <inheritdoc />
  60. public float? YarnBetaFast { get; set; }
  61. /// <inheritdoc />
  62. public float? YarnBetaSlow { get; set; }
  63. /// <inheritdoc />
  64. public uint? YarnOriginalContext { get; set; }
  65. /// <inheritdoc />
  66. public RopeScalingType? YarnScalingType { get; set; }
  67. /// <inheritdoc />
  68. public GGMLType? TypeK { get; set; }
  69. /// <inheritdoc />
  70. public GGMLType? TypeV { get; set; }
  71. /// <inheritdoc />
  72. public bool NoKqvOffload { get; set; }
  73. /// <inheritdoc />
  74. public Encoding Encoding { get; set; } = Encoding.UTF8;
  75. /// <inheritdoc />
  76. public bool VocabOnly { get; set; }
  77. /// <inheritdoc />
  78. public float DefragThreshold { get; set; }
  79. /// <inheritdoc />
  80. public LLamaPoolingType PoolingType { get; set; }
  81. }
  82. }