May 2024 Binary Update (Take 2)

LLama.Examples/Examples/BatchedExecutorGuidance.cs

@@ -79,7 +79,7 @@ await AnsiConsole
                     guidance.Prompt(g);
 
                     // Early exit if we reach the natural end of the guided sentence
-                    if (g == model.Tokens.EOS)
+                    if (model.Tokens.IsEndOfGeneration(g))
                         break;
 
                     // Update progress bar

LLama.Web/Common/ModelOptions.cs

@@ -29,12 +29,14 @@ public class ModelOptions
         /// <inheritdoc />
         public int GpuLayerCount { get; set; } = 20;
 
-        public uint SeqMax { get; }
+        /// <inheritdoc />
+        public uint SeqMax { get; set; }
 
         /// <inheritdoc />
         public uint Seed { get; set; } = 1686349486;
 
-        public bool Embeddings { get; }
+        /// <inheritdoc />
+        public bool Embeddings { get; set; }
 
         /// <inheritdoc />
         public bool UseMemorymap { get; set; } = true;
@@ -102,6 +104,9 @@ public class ModelOptions
         /// <inheritdoc />
         public bool NoKqvOffload { get; set; }
 
+        /// <inheritdoc />
+        public bool FlashAttention { get; set; }
+
         /// <inheritdoc />
         public Encoding Encoding { get; set; } = Encoding.UTF8;
 

LLama/Abstractions/IContextParams.cs

@@ -108,6 +108,11 @@ public interface IContextParams
     /// </summary>
     bool NoKqvOffload { get; }
 
+    /// <summary>
+    /// Whether to use flash attention
+    /// </summary>
+    bool FlashAttention { get; }
+
     /// <summary>
     /// defragment the KV cache if holes/size &gt; defrag_threshold, Set to &lt; 0 to disable (default)
     /// </summary>

LLama/Abstractions/IModelParams.cs

@@ -4,6 +4,7 @@
 using System.Collections.Generic;
 using System.ComponentModel;
 using System.Linq;
+using System.Text;
 using System.Text.Json;
 using System.Text.Json.Serialization;
 using LLama.Native;
@@ -241,6 +242,7 @@ public sealed record MetadataOverride
         private readonly int _valueInt;
         private readonly float _valueFloat;
         private readonly bool _valueBool;
+        private readonly byte[]? _valueString;
 
         /// <summary>
         /// Create a new override for an int key
@@ -278,6 +280,21 @@ public MetadataOverride(string key, bool value)
             Type = LLamaModelKvOverrideType.Bool;
         }
 
+        /// <summary>
+        /// Create a new override for a string key
+        /// </summary>
+        /// <param name="key"></param>
+        /// <param name="value"></param>
+        public MetadataOverride(string key, string value)
+        {
+            Key = key;
+            _valueString = Encoding.UTF8.GetBytes(value); 
+            Type = LLamaModelKvOverrideType.String;
+
+            if (_valueString.Length > 128)
+                throw new ArgumentException("Value string is too long, must be < 128 UTF8 bytes", nameof(value));
+        }
+
         internal void WriteValue(ref LLamaModelMetadataOverride dest)
         {
             switch (Type)
@@ -291,6 +308,13 @@ internal void WriteValue(ref LLamaModelMetadataOverride dest)
                 case LLamaModelKvOverrideType.Bool:
                     dest.BoolValue = _valueBool ? -1L : 0;
                     break;
+                case LLamaModelKvOverrideType.String:
+                    unsafe
+                    {
+                        fixed (byte* strValPtr = dest.StringValue)
+                            new Span<byte>(_valueString!).CopyTo(new Span<byte>(strValPtr, 128));
+                    }
+                    break;
                 default:
                     throw new InvalidEnumArgumentException($"Unknown {nameof(LLamaModelKvOverrideType)} value: {Type}");
             }

LLama/Common/ModelParams.cs

@@ -99,6 +99,9 @@ public record ModelParams
         /// <inheritdoc />
         public bool NoKqvOffload { get; set; }
 
+        /// <inheritdoc />
+        public bool FlashAttention { get; set; }
+
         /// <inheritdoc />
         public float DefragThreshold { get; set; }
 

LLama/Extensions/IContextParamsExtensions.cs

@@ -50,6 +50,7 @@ public static void ToLlamaContextParams(this IContextParams @params, out LLamaCo
             result.type_k = @params.TypeK ?? GGMLType.GGML_TYPE_F16;
             result.type_k = @params.TypeV ?? GGMLType.GGML_TYPE_F16;
             result.offload_kqv = !@params.NoKqvOffload;
+            result.flash_attention = @params.FlashAttention;
             result.llama_pooling_type = @params.PoolingType;
 
             result.n_threads = Threads(@params.Threads);

LLama/LLamaStatelessExecutor.cs

@@ -4,6 +4,7 @@
 using System.Collections.Generic;
 using System.Linq;
 using System.Runtime.CompilerServices;
+using System.Text;
 using System.Threading;
 using LLama.Exceptions;
 using LLama.Native;
@@ -123,8 +124,8 @@ public async IAsyncEnumerable<string> InferAsync(string prompt, IInferenceParams
                     );
                 }
 
-                // Check if this is the EOS token
-                if (id == _weights.Tokens.EOS)
+                // Check if this token should end generation
+                if (_weights.Tokens.IsEndOfGeneration(id))
                     break;
 
                 // Decode this token into text

LLama/Native/LLamaContextParams.cs

@@ -151,6 +151,16 @@ public bool offload_kqv
         }
         private sbyte _offload_kqv;
 
+        /// <summary>
+        /// whether to use flash attention
+        /// </summary>
+        public bool flash_attention
+        {
+            readonly get => Convert.ToBoolean(_flash_attention);
+            set => _flash_attention = Convert.ToSByte(value);
+        }
+        private sbyte _flash_attention;
+
         //todo: implement abort callback support
         /// <summary>
         /// ggml_abort_callback

LLama/Native/LLamaFtype.cs

@@ -171,6 +171,11 @@ public enum LLamaFtype
         /// </summary>
         LLAMA_FTYPE_MOSTLY_IQ1_M = 31,
 
+        /// <summary>
+        /// except 1d tensors
+        /// </summary>
+        LLAMA_FTYPE_MOSTLY_BF16 = 32,
+
         /// <summary>
         /// File type was not specified
         /// </summary>

LLama/Native/LLamaModelMetadataOverride.cs

@@ -43,6 +43,12 @@ public unsafe struct LLamaModelMetadataOverride
     /// </summary>
     [FieldOffset(136)]
     public long BoolValue;
+
+    /// <summary>
+    /// Value, **must** only be used if Tag == String
+    /// </summary>
+    [FieldOffset(136)]
+    public fixed byte StringValue[128];
 }
 
 /// <summary>
@@ -65,4 +71,9 @@ public enum LLamaModelKvOverrideType
     /// Overriding a bool value
     /// </summary>
     Bool = 2,
+
+    /// <summary>
+    /// Overriding a string value
+    /// </summary>
+    String = 3,
 }

LLama/Native/LLamaModelParams.cs

@@ -81,6 +81,16 @@ public bool use_mlock
         }
         private sbyte _use_mlock;
 
+        /// <summary>
+        /// validate model tensor data
+        /// </summary>
+        public bool check_tensors
+        {
+            readonly get => Convert.ToBoolean(_check_tensors);
+            set => _check_tensors = Convert.ToSByte(value);
+        }
+        private sbyte _check_tensors;
+
         /// <summary>
         /// Create a LLamaModelParams with default values
         /// </summary>

LLama/Native/LLamaModelQuantizeParams.cs

@@ -70,6 +70,16 @@ public bool pure
         }
         private sbyte _pure;
 
+        /// <summary>
+        /// quantize to the same number of shards
+        /// </summary>
+        public bool keep_split
+        {
+            get => Convert.ToBoolean(_keep_split);
+            set => _keep_split = Convert.ToSByte(value);
+        }
+        private sbyte _keep_split;
+
         /// <summary>
         /// pointer to importance matrix data
         /// </summary>

LLama/Native/LLamaVocabPreType.cs

@@ -0,0 +1,17 @@
+namespace LLama.Native;
+
+/// <summary>
+/// 
+/// </summary>
+/// <remarks>llama_vocab_pre_type</remarks>
+internal enum LLamaVocabPreType
+{
+    LLAMA_VOCAB_PRE_TYPE_DEFAULT = 0,
+    LLAMA_VOCAB_PRE_TYPE_LLAMA3 = 1,
+    LLAMA_VOCAB_PRE_TYPE_DEEPSEEK_LLM = 2,
+    LLAMA_VOCAB_PRE_TYPE_DEEPSEEK_CODER = 3,
+    LLAMA_VOCAB_PRE_TYPE_FALCON = 4,
+    LLAMA_VOCAB_PRE_TYPE_MPT = 5,
+    LLAMA_VOCAB_PRE_TYPE_STARCODER = 6,
+    LLAMA_VOCAB_PRE_TYPE_GPT2 = 7,
+}

LLama/Native/NativeApi.LLava.cs

@@ -13,6 +13,7 @@ public static unsafe partial class NativeApi
     /// <param name="ctxClip">Llava Model</param>
     /// <returns>True if validate successfully</returns>
     [DllImport(llavaLibraryName, EntryPoint = "llava_validate_embed_size", CallingConvention = CallingConvention.Cdecl)]
+    [return: MarshalAs(UnmanagedType.U1)]
     public static extern bool llava_validate_embed_size( SafeLLamaContextHandle ctxLlama, SafeLlavaModelHandle ctxClip);
 
     /// <summary>
@@ -56,7 +57,7 @@ SafeLlavaImageEmbedHandle llava_image_embed_make_with_filename(SafeLlavaModelHan
     /// <param name="embed">Embedding handle</param>
     /// <returns>True on success</returns>
     [DllImport(llavaLibraryName, EntryPoint = "llava_eval_image_embed", CallingConvention = CallingConvention.Cdecl)]
-    public static extern bool llava_eval_image_embed(SafeLLamaContextHandle ctx_llama, SafeLlavaImageEmbedHandle embed,
-        int n_batch, ref int n_past);
+    [return: MarshalAs(UnmanagedType.U1)]
+    public static extern bool llava_eval_image_embed(SafeLLamaContextHandle ctx_llama, SafeLlavaImageEmbedHandle embed, int n_batch, ref int n_past);
 
 }

LLama/Native/NativeApi.Sampling.cs

@@ -176,7 +176,7 @@ public static void llama_sample_apply_guidance(SafeLLamaContextHandle ctx, Span<
         public static extern LLamaToken llama_sample_token_greedy(SafeLLamaContextHandle ctx, ref LLamaTokenDataArrayNative candidates);
 
         /// <summary>
-        /// Randomly selects a token from the candidates based on their probabilities.
+        /// Randomly selects a token from the candidates based on their probabilities using the RNG of ctx.
         /// </summary>
         /// <param name="ctx"></param>
         /// <param name="candidates">Pointer to LLamaTokenDataArray</param>

LLama/Native/NativeApi.cs

@@ -34,20 +34,23 @@ public static void llama_empty_call()
         /// </summary>
         /// <returns></returns>
         [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl)]
+        [return: MarshalAs(UnmanagedType.U1)]
         public static extern bool llama_supports_mmap();
 
         /// <summary>
         /// Check if memory locking is supported
         /// </summary>
         /// <returns></returns>
         [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl)]
+        [return: MarshalAs(UnmanagedType.U1)]
         public static extern bool llama_supports_mlock();
 
         /// <summary>
         /// Check if GPU offload is supported
         /// </summary>
         /// <returns></returns>
         [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl)]
+        [return: MarshalAs(UnmanagedType.U1)]
         public static extern bool llama_supports_gpu_offload();
 
         /// <summary>
@@ -77,6 +80,7 @@ public static void llama_empty_call()
         /// <param name="n_token_count_out"></param>
         /// <returns></returns>
         [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl)]
+        [return: MarshalAs(UnmanagedType.U1)]
         public static extern bool llama_state_load_file(SafeLLamaContextHandle ctx, string path_session, LLamaToken[] tokens_out, ulong n_token_capacity, out ulong n_token_count_out);
 
         /// <summary>
@@ -88,6 +92,7 @@ public static void llama_empty_call()
         /// <param name="n_token_count"></param>
         /// <returns></returns>
         [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl)]
+        [return: MarshalAs(UnmanagedType.U1)]
         public static extern bool llama_state_save_file(SafeLLamaContextHandle ctx, string path_session, LLamaToken[] tokens, ulong n_token_count);
 
         [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl)]
@@ -133,6 +138,14 @@ public static void llama_empty_call()
         [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl)]
         public static extern uint llama_n_seq_max(SafeLLamaContextHandle ctx);
 
+        /// <summary>
+        /// Get the pooling type for this context
+        /// </summary>
+        /// <param name="ctx"></param>
+        /// <returns></returns>
+        [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl)]
+        public static extern LLamaPoolingType llama_pooling_type(SafeLLamaContextHandle ctx);
+
         /// <summary>
         /// Get the embeddings for the a specific sequence.
         /// Equivalent to: llama_get_embeddings(ctx) + ctx->output_ids[i]*n_embd
@@ -218,19 +231,20 @@ public static void llama_empty_call()
         /// <param name="model"></param>
         /// <param name="llamaToken"></param>
         /// <param name="buffer">buffer to write string into</param>
+        /// <param name="special">If true, special tokens are rendered in the output</param>
         /// <returns>The length written, or if the buffer is too small a negative that indicates the length required</returns>
-        public static int llama_token_to_piece(SafeLlamaModelHandle model, LLamaToken llamaToken, Span<byte> buffer)
+        public static int llama_token_to_piece(SafeLlamaModelHandle model, LLamaToken llamaToken, Span<byte> buffer, bool special)
         {
             unsafe
             {
                 fixed (byte* bufferPtr = buffer)
                 {
-                    return llama_token_to_piece_native(model, llamaToken, bufferPtr, buffer.Length);
+                    return llama_token_to_piece_native(model, llamaToken, bufferPtr, buffer.Length, special);
                 }
             }
 
             [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl, EntryPoint = "llama_token_to_piece")]
-            static extern unsafe int llama_token_to_piece_native(SafeLlamaModelHandle model, LLamaToken llamaToken, byte* buffer, int length);
+            static extern unsafe int llama_token_to_piece_native(SafeLlamaModelHandle model, LLamaToken llamaToken, byte* buffer, int length, bool special);
         }
 
         /// <summary>
@@ -260,7 +274,7 @@ public static void llama_log_set(NativeLogConfig.LLamaLogCallback logCallback)
         }
 
         /// <summary>
-        /// Clear the KV cache
+        /// Clear the KV cache. Both cell info is erased and KV data is zeroed
         /// </summary>
         /// <param name="ctx"></param>
         [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl)]
@@ -275,6 +289,7 @@ public static void llama_log_set(NativeLogConfig.LLamaLogCallback logCallback)
         /// <param name="p1"></param>
         /// <returns>Returns false if a partial sequence cannot be removed. Removing a whole sequence never fails</returns>
         [DllImport(libraryName, CallingConvention = CallingConvention.Cdecl)]
+        [return: MarshalAs(UnmanagedType.U1)]
         public static extern bool llama_kv_cache_seq_rm(SafeLLamaContextHandle ctx, LLamaSeqId seq, LLamaPos p0, LLamaPos p1);
 
         /// <summary>