@fugood/llama.node 0.0.1-alpha.3 → 0.1.0

package/src/llama.cpp/examples/llama-bench/llama-bench.cpp

@@ -174,9 +174,11 @@ struct cmd_params {
     std::vector<llama_split_mode> split_mode;
     std::vector<int> main_gpu;
     std::vector<bool> no_kv_offload;
+    std::vector<bool> flash_attn;
     std::vector<std::vector<float>> tensor_split;
     std::vector<bool> use_mmap;
     std::vector<bool> embeddings;
+    ggml_numa_strategy numa;
     int reps;
     bool verbose;
     output_formats output_format;
@@ -195,9 +197,11 @@ static const cmd_params cmd_params_defaults = {
     /* split_mode    */ {LLAMA_SPLIT_MODE_LAYER},
     /* main_gpu      */ {0},
     /* no_kv_offload */ {false},
+    /* flash_attn    */ {false},
     /* tensor_split  */ {std::vector<float>(llama_max_devices(), 0.0f)},
     /* use_mmap      */ {true},
     /* embeddings    */ {false},
+    /* numa          */ GGML_NUMA_STRATEGY_DISABLED,
     /* reps          */ 5,
     /* verbose       */ false,
     /* output_format */ MARKDOWN
@@ -220,7 +224,9 @@ static void print_usage(int /* argc */, char ** argv) {
     printf("  -sm, --split-mode <none|layer|row>  (default: %s)\n", join(transform_to_str(cmd_params_defaults.split_mode, split_mode_str), ",").c_str());
     printf("  -mg, --main-gpu <i>                 (default: %s)\n", join(cmd_params_defaults.main_gpu, ",").c_str());
     printf("  -nkvo, --no-kv-offload <0|1>        (default: %s)\n", join(cmd_params_defaults.no_kv_offload, ",").c_str());
+    printf("  -fa, --flash-attn <0|1>             (default: %s)\n", join(cmd_params_defaults.flash_attn, ",").c_str());
     printf("  -mmp, --mmap <0|1>                  (default: %s)\n", join(cmd_params_defaults.use_mmap, ",").c_str());
+    printf("  --numa <distribute|isolate|numactl> (default: disabled)\n");
     printf("  -embd, --embeddings <0|1>           (default: %s)\n", join(cmd_params_defaults.embeddings, ",").c_str());
     printf("  -ts, --tensor-split <ts0/ts1/..>    (default: 0)\n");
     printf("  -r, --repetitions <n>               (default: %d)\n", cmd_params_defaults.reps);
@@ -393,6 +399,24 @@ static cmd_params parse_cmd_params(int argc, char ** argv) {
             }
             auto p = split<bool>(argv[i], split_delim);
             params.no_kv_offload.insert(params.no_kv_offload.end(), p.begin(), p.end());
+        } else if (arg == "--numa") {
+            if (++i >= argc) {
+                invalid_param = true;
+                break;
+            } else {
+                std::string value(argv[i]);
+                /**/ if (value == "distribute" || value == "" ) { params.numa = GGML_NUMA_STRATEGY_DISTRIBUTE; }
+                else if (value == "isolate")                    { params.numa = GGML_NUMA_STRATEGY_ISOLATE; }
+                else if (value == "numactl")                    { params.numa = GGML_NUMA_STRATEGY_NUMACTL; }
+                else { invalid_param = true; break; }
+            }
+        } else if (arg == "-fa" || arg == "--flash-attn") {
+            if (++i >= argc) {
+                invalid_param = true;
+                break;
+            }
+            auto p = split<bool>(argv[i], split_delim);
+            params.flash_attn.insert(params.flash_attn.end(), p.begin(), p.end());
         } else if (arg == "-mmp" || arg == "--mmap") {
             if (++i >= argc) {
                 invalid_param = true;
@@ -477,6 +501,7 @@ static cmd_params parse_cmd_params(int argc, char ** argv) {
     if (params.split_mode.empty())   { params.split_mode = cmd_params_defaults.split_mode; }
     if (params.main_gpu.empty())     { params.main_gpu = cmd_params_defaults.main_gpu; }
     if (params.no_kv_offload.empty()){ params.no_kv_offload = cmd_params_defaults.no_kv_offload; }
+    if (params.flash_attn.empty())   { params.flash_attn = cmd_params_defaults.flash_attn; }
     if (params.tensor_split.empty()) { params.tensor_split = cmd_params_defaults.tensor_split; }
     if (params.use_mmap.empty())     { params.use_mmap = cmd_params_defaults.use_mmap; }
     if (params.embeddings.empty())   { params.embeddings = cmd_params_defaults.embeddings; }
@@ -498,6 +523,7 @@ struct cmd_params_instance {
     llama_split_mode split_mode;
     int main_gpu;
     bool no_kv_offload;
+    bool flash_attn;
     std::vector<float> tensor_split;
     bool use_mmap;
     bool embeddings;
@@ -532,6 +558,7 @@ struct cmd_params_instance {
         cparams.type_k = type_k;
         cparams.type_v = type_v;
         cparams.offload_kqv = !no_kv_offload;
+        cparams.flash_attn = flash_attn;
         cparams.embeddings = embeddings;
 
         return cparams;
@@ -554,6 +581,7 @@ static std::vector<cmd_params_instance> get_cmd_params_instances(const cmd_param
     for (const auto & tk : params.type_k)
     for (const auto & tv : params.type_v)
     for (const auto & nkvo : params.no_kv_offload)
+    for (const auto & fa : params.flash_attn)
     for (const auto & nt : params.n_threads) {
         for (const auto & n_prompt : params.n_prompt) {
             if (n_prompt == 0) {
@@ -572,6 +600,7 @@ static std::vector<cmd_params_instance> get_cmd_params_instances(const cmd_param
                 /* .split_mode   = */ sm,
                 /* .main_gpu     = */ mg,
                 /* .no_kv_offload= */ nkvo,
+                /* .flash_attn   = */ fa,
                 /* .tensor_split = */ ts,
                 /* .use_mmap     = */ mmp,
                 /* .embeddings   = */ embd,
@@ -596,6 +625,7 @@ static std::vector<cmd_params_instance> get_cmd_params_instances(const cmd_param
                 /* .split_mode   = */ sm,
                 /* .main_gpu     = */ mg,
                 /* .no_kv_offload= */ nkvo,
+                /* .flash_attn   = */ fa,
                 /* .tensor_split = */ ts,
                 /* .use_mmap     = */ mmp,
                 /* .embeddings   = */ embd,
@@ -633,6 +663,7 @@ struct test {
     llama_split_mode split_mode;
     int main_gpu;
     bool no_kv_offload;
+    bool flash_attn;
     std::vector<float> tensor_split;
     bool use_mmap;
     bool embeddings;
@@ -657,6 +688,7 @@ struct test {
         split_mode = inst.split_mode;
         main_gpu = inst.main_gpu;
         no_kv_offload = inst.no_kv_offload;
+        flash_attn = inst.flash_attn;
         tensor_split = inst.tensor_split;
         use_mmap = inst.use_mmap;
         embeddings = inst.embeddings;
@@ -731,7 +763,7 @@ struct test {
             "n_batch", "n_ubatch",
             "n_threads", "type_k", "type_v",
             "n_gpu_layers", "split_mode",
-            "main_gpu", "no_kv_offload",
+            "main_gpu", "no_kv_offload", "flash_attn",
             "tensor_split", "use_mmap", "embeddings",
             "n_prompt", "n_gen", "test_time",
             "avg_ns", "stddev_ns",
@@ -753,7 +785,7 @@ struct test {
         }
         if (field == "cuda" || field == "opencl"  || field == "vulkan" || field == "kompute" || field == "metal" ||
             field == "gpu_blas" || field == "blas" || field == "sycl" ||field == "f16_kv" || field == "no_kv_offload" ||
-            field == "use_mmap" || field == "embeddings") {
+            field == "flash_attn" || field == "use_mmap" || field == "embeddings") {
             return BOOL;
         }
         if (field == "avg_ts" || field == "stddev_ts") {
@@ -787,7 +819,7 @@ struct test {
             std::to_string(n_batch), std::to_string(n_ubatch),
             std::to_string(n_threads), ggml_type_name(type_k), ggml_type_name(type_v),
             std::to_string(n_gpu_layers), split_mode_str(split_mode),
-            std::to_string(main_gpu), std::to_string(no_kv_offload),
+            std::to_string(main_gpu), std::to_string(no_kv_offload), std::to_string(flash_attn),
             tensor_split_str, std::to_string(use_mmap), std::to_string(embeddings),
             std::to_string(n_prompt), std::to_string(n_gen), test_time,
             std::to_string(avg_ns()), std::to_string(stdev_ns()),
@@ -955,6 +987,9 @@ struct markdown_printer : public printer {
         if (field == "no_kv_offload") {
             return "nkvo";
         }
+        if (field == "flash_attn") {
+            return "fa";
+        }
         if (field == "use_mmap") {
             return "mmap";
         }
@@ -1001,6 +1036,9 @@ struct markdown_printer : public printer {
         if (params.no_kv_offload.size() > 1 || params.no_kv_offload != cmd_params_defaults.no_kv_offload) {
             fields.emplace_back("no_kv_offload");
         }
+        if (params.flash_attn.size() > 1 || params.flash_attn != cmd_params_defaults.flash_attn) {
+            fields.emplace_back("flash_attn");
+        }
         if (params.tensor_split.size() > 1 || params.tensor_split != cmd_params_defaults.tensor_split) {
             fields.emplace_back("tensor_split");
         }
@@ -1191,6 +1229,7 @@ int main(int argc, char ** argv) {
         llama_log_set(llama_null_log_callback, NULL);
     }
     llama_backend_init();
+    llama_numa_init(params.numa);
 
     // initialize printer
     std::unique_ptr<printer> p;

package/src/llama.cpp/examples/llava/clip.cpp

@@ -104,6 +104,7 @@ static std::string format(const char * fmt, ...) {
 #define TN_POS_EMBD        "%s.position_embd.weight"
 #define TN_CLASS_EMBD      "v.class_embd"
 #define TN_PATCH_EMBD      "v.patch_embd.weight"
+#define TN_PATCH_BIAS      "v.patch_embd.bias"
 #define TN_ATTN_K          "%s.blk.%d.attn_k.%s"
 #define TN_ATTN_Q          "%s.blk.%d.attn_q.%s"
 #define TN_ATTN_V          "%s.blk.%d.attn_v.%s"
@@ -425,6 +426,7 @@ struct clip_vision_model {
     // embeddings
     struct ggml_tensor * class_embedding;
     struct ggml_tensor * patch_embeddings;
+    struct ggml_tensor * patch_bias;
     struct ggml_tensor * position_embeddings;
 
     struct ggml_tensor * pre_ln_w;
@@ -501,6 +503,11 @@ struct clip_ctx {
     bool use_gelu = false;
     int32_t ftype = 1;
 
+    bool has_class_embedding = true;
+    bool has_pre_norm = true;
+    bool has_post_norm = false;
+    bool has_patch_bias = false;
+
     struct gguf_context * ctx_gguf;
     struct ggml_context * ctx_data;
 
@@ -526,7 +533,7 @@ static ggml_cgraph * clip_image_build_graph(clip_ctx * ctx, const clip_image_f32
     const int patch_size           = hparams.patch_size;
     const int num_patches          = ((image_size / patch_size) * (image_size / patch_size));
     const int num_patches_per_side = image_size / patch_size; GGML_UNUSED(num_patches_per_side);
-    const int num_positions        = num_patches + 1;
+    const int num_positions        = num_patches + (ctx->has_class_embedding ? 1 : 0);
     const int hidden_size          = hparams.hidden_size;
     const int n_head               = hparams.n_head;
     const int d_head               = hidden_size / n_head;
@@ -557,16 +564,23 @@ static ggml_cgraph * clip_image_build_graph(clip_ctx * ctx, const clip_image_f32
     inp = ggml_reshape_3d(ctx0, inp, num_patches, hidden_size, batch_size);
     inp = ggml_cont(ctx0, ggml_permute(ctx0, inp, 1, 0, 2, 3));
 
-    // concat class_embeddings and patch_embeddings
-    struct ggml_tensor * embeddings = ggml_new_tensor_3d(ctx0, GGML_TYPE_F32, hidden_size, num_positions, batch_size);
-    ggml_set_name(embeddings, "embeddings");
-    ggml_set_input(embeddings);
+    if (ctx->has_patch_bias) {
+        // inp = ggml_add(ctx0, inp, ggml_repeat(ctx0, model.patch_bias, inp));
+        inp = ggml_add(ctx0, inp, model.patch_bias);
+    }
 
-    embeddings = ggml_acc(ctx0, embeddings, model.class_embedding,
-            embeddings->nb[1], embeddings->nb[2], embeddings->nb[3], 0);
+    // concat class_embeddings and patch_embeddings
+    struct ggml_tensor * embeddings = inp;
+    if (ctx->has_class_embedding) {
+        embeddings = ggml_new_tensor_3d(ctx0, GGML_TYPE_F32, hidden_size, num_positions, batch_size);
+        ggml_set_name(embeddings, "embeddings");
+        ggml_set_input(embeddings);
+        embeddings = ggml_acc(ctx0, embeddings, model.class_embedding,
+                embeddings->nb[1], embeddings->nb[2], embeddings->nb[3], 0);
+        embeddings = ggml_acc(ctx0, embeddings, inp,
+                embeddings->nb[1], embeddings->nb[2], embeddings->nb[3], model.class_embedding->nb[1]);
+    }
 
-    embeddings = ggml_acc(ctx0, embeddings, inp,
-            embeddings->nb[1], embeddings->nb[2], embeddings->nb[3], model.class_embedding->nb[1]);
 
     struct ggml_tensor * positions = ggml_new_tensor_1d(ctx0, GGML_TYPE_I32, num_positions);
     ggml_set_name(positions, "positions");
@@ -576,7 +590,7 @@ static ggml_cgraph * clip_image_build_graph(clip_ctx * ctx, const clip_image_f32
         ggml_add(ctx0, embeddings, ggml_get_rows(ctx0, model.position_embeddings, positions));
 
     // pre-layernorm
-    {
+    if (ctx->has_pre_norm) {
         embeddings = ggml_norm(ctx0, embeddings, eps);
         ggml_set_name(embeddings, "pre_ln");
 
@@ -664,6 +678,14 @@ static ggml_cgraph * clip_image_build_graph(clip_ctx * ctx, const clip_image_f32
         embeddings = cur;
     }
 
+    // post-layernorm
+    if (ctx->has_post_norm) {
+        embeddings = ggml_norm(ctx0, embeddings, eps);
+        ggml_set_name(embeddings, "post_ln");
+
+        embeddings = ggml_add(ctx0, ggml_mul(ctx0, embeddings, model.post_ln_w), model.post_ln_b);
+    }
+
     // llava projector
     {
         embeddings = ggml_reshape_2d(ctx0, embeddings, embeddings->ne[0], embeddings->ne[1]);
@@ -1148,12 +1170,39 @@ struct clip_ctx * clip_model_load(const char * fname, const int verbosity = 1) {
 
         }
 
+        try {
+            vision_model.class_embedding  = get_tensor(new_clip->ctx_data, TN_CLASS_EMBD);
+            new_clip->has_class_embedding = true;
+        } catch (const std::exception& e) {
+            new_clip->has_class_embedding = false;
+        }
+
+        try {
+            vision_model.pre_ln_w  = get_tensor(new_clip->ctx_data, format(TN_LN_PRE, "v", "weight"));
+            vision_model.pre_ln_b  = get_tensor(new_clip->ctx_data, format(TN_LN_PRE, "v", "bias"));
+            new_clip->has_pre_norm = true;
+        } catch (std::exception & e) {
+            new_clip->has_pre_norm = false;
+        }
+
+        try {
+            vision_model.post_ln_w  = get_tensor(new_clip->ctx_data, format(TN_LN_POST, "v", "weight"));
+            vision_model.post_ln_b  = get_tensor(new_clip->ctx_data, format(TN_LN_POST, "v", "bias"));
+            new_clip->has_post_norm = true;
+        } catch (std::exception & e) {
+            new_clip->has_post_norm = false;
+        }
+
+        try {
+            vision_model.patch_bias = get_tensor(new_clip->ctx_data, TN_PATCH_BIAS);
+            new_clip->has_patch_bias = true;
+        } catch (std::exception & e) {
+            new_clip->has_patch_bias = false;
+        }
+
         try {
             vision_model.patch_embeddings    = get_tensor(new_clip->ctx_data, TN_PATCH_EMBD);
-            vision_model.class_embedding     = get_tensor(new_clip->ctx_data, TN_CLASS_EMBD);
             vision_model.position_embeddings = get_tensor(new_clip->ctx_data, format(TN_POS_EMBD, "v"));
-            vision_model.pre_ln_w            = get_tensor(new_clip->ctx_data, format(TN_LN_PRE, "v", "weight"));
-            vision_model.pre_ln_b            = get_tensor(new_clip->ctx_data, format(TN_LN_PRE, "v", "bias"));
         } catch(const std::exception& e) {
             LOG_TEE("%s: failed to load vision model tensors\n", __func__);
         }
@@ -1325,7 +1374,7 @@ bool clip_image_load_from_bytes(const unsigned char * bytes, size_t bytes_length
 }
 
 // Linear interpolation between two points
-inline float lerp(float s, float e, float t) {
+inline float clip_lerp(float s, float e, float t) {
     return s + (e - s) * t;
 }
 // Bilinear resize function
@@ -1347,17 +1396,17 @@ static void bilinear_resize(const clip_image_u8& src, clip_image_u8& dst, int ta
             float y_lerp = py - y_floor;
 
             for (int c = 0; c < 3; c++) {
-                float top = lerp(
+                float top = clip_lerp(
                     static_cast<float>(src.buf[3 * (y_floor * src.nx + x_floor) + c]),
                     static_cast<float>(src.buf[3 * (y_floor * src.nx + (x_floor + 1)) + c]),
                     x_lerp
                 );
-                float bottom = lerp(
+                float bottom = clip_lerp(
                     static_cast<float>(src.buf[3 * ((y_floor + 1) * src.nx + x_floor) + c]),
                     static_cast<float>(src.buf[3 * ((y_floor + 1) * src.nx + (x_floor + 1)) + c]),
                     x_lerp
                 );
-                dst.buf[3 * (y * target_width + x) + c] = static_cast<uint8_t>(lerp(top, bottom, y_lerp));
+                dst.buf[3 * (y * target_width + x) + c] = static_cast<uint8_t>(clip_lerp(top, bottom, y_lerp));
             }
         }
     }
@@ -1797,7 +1846,7 @@ bool clip_image_batch_encode(clip_ctx * ctx, const int n_threads, const clip_ima
     const int image_size    = hparams.image_size;
     const int patch_size    = hparams.patch_size;
     const int num_patches   = ((image_size / patch_size) * (image_size / patch_size));
-    const int num_positions = num_patches + 1;
+    const int num_positions = num_patches + (ctx->has_class_embedding ? 1 : 0);
 
     {
         struct ggml_tensor * inp_raw = ggml_graph_get_tensor(gf, "inp_raw");
@@ -1825,12 +1874,14 @@ bool clip_image_batch_encode(clip_ctx * ctx, const int n_threads, const clip_ima
     }
 
     {
-        struct ggml_tensor * embeddings = ggml_graph_get_tensor(gf, "embeddings");
+        if (ctx->has_class_embedding) {
+            struct ggml_tensor * embeddings = ggml_graph_get_tensor(gf, "embeddings");
 
-        void* zero_mem = malloc(ggml_nbytes(embeddings));
-        memset(zero_mem, 0, ggml_nbytes(embeddings));
-        ggml_backend_tensor_set(embeddings, zero_mem, 0, ggml_nbytes(embeddings));
-        free(zero_mem);
+            void* zero_mem = malloc(ggml_nbytes(embeddings));
+            memset(zero_mem, 0, ggml_nbytes(embeddings));
+            ggml_backend_tensor_set(embeddings, zero_mem, 0, ggml_nbytes(embeddings));
+            free(zero_mem);
+        }
     }
 
     {

package/src/llama.cpp/examples/llava/llava-cli.cpp

@@ -113,11 +113,11 @@ struct llava_context {
 };
 
 static void show_additional_info(int /*argc*/, char ** argv) {
-    LOG_TEE("\n example usage: %s -m <llava-v1.5-7b/ggml-model-q5_k.gguf> --mmproj <llava-v1.5-7b/mmproj-model-f16.gguf> --image <path/to/an/image.jpg> [--temp 0.1] [-p \"describe the image in detail.\"]\n", argv[0]);
+    LOG_TEE("\n example usage: %s -m <llava-v1.5-7b/ggml-model-q5_k.gguf> --mmproj <llava-v1.5-7b/mmproj-model-f16.gguf> --image <path/to/an/image.jpg> --image <path/to/another/image.jpg> [--temp 0.1] [-p \"describe the image in detail.\"]\n", argv[0]);
     LOG_TEE("  note: a lower temperature value like 0.1 is recommended for better quality.\n");
 }
 
-static struct llava_image_embed * load_image(llava_context * ctx_llava, gpt_params * params) {
+static struct llava_image_embed * load_image(llava_context * ctx_llava, gpt_params * params, const std::string & fname) {
 
     // load and preprocess the image
     llava_image_embed * embed = NULL;
@@ -133,9 +133,9 @@ static struct llava_image_embed * load_image(llava_context * ctx_llava, gpt_para
         }
         params->prompt = remove_image_from_prompt(prompt);
     } else {
-        embed = llava_image_embed_make_with_filename(ctx_llava->ctx_clip, params->n_threads, params->image.c_str());
+        embed = llava_image_embed_make_with_filename(ctx_llava->ctx_clip, params->n_threads, fname.c_str());
         if (!embed) {
-            LOG_TEE("%s: is %s really an image file?\n", __func__, params->image.c_str());
+            fprintf(stderr, "%s: is %s really an image file?\n", __func__, fname.c_str());
             return NULL;
         }
     }
@@ -207,17 +207,7 @@ static void process_prompt(struct llava_context * ctx_llava, struct llava_image_
     printf("\n");
 }
 
-
-static struct llava_context * llava_init(gpt_params * params) {
-    const char * clip_path = params->mmproj.c_str();
-
-    auto prompt = params->prompt;
-    if (prompt.empty()) {
-        prompt = "describe the image in detail.";
-    }
-
-    auto ctx_clip = clip_model_load(clip_path, /*verbosity=*/ 1);
-
+static struct llama_model * llava_init(gpt_params * params) {
     llama_backend_init();
     llama_numa_init(params->numa);
 
@@ -228,6 +218,19 @@ static struct llava_context * llava_init(gpt_params * params) {
         LOG_TEE("%s: error: unable to load model\n" , __func__);
         return NULL;
     }
+    return model;
+}
+
+static struct llava_context * llava_init_context(gpt_params * params, llama_model * model) {
+    const char * clip_path = params->mmproj.c_str();
+
+    auto prompt = params->prompt;
+    if (prompt.empty()) {
+        prompt = "describe the image in detail.";
+    }
+
+    auto ctx_clip = clip_model_load(clip_path, /*verbosity=*/ 1);
+
 
     llama_context_params ctx_params = llama_context_params_from_gpt_params(*params);
     ctx_params.n_ctx           = params->n_ctx < 2048 ? 2048 : params->n_ctx; // we need a longer context size to process image embeddings
@@ -286,24 +289,30 @@ int main(int argc, char ** argv) {
         show_additional_info(argc, argv);
         return 1;
     }
-
-    auto ctx_llava = llava_init(&params);
-    if (ctx_llava == NULL) {
-        LOG_TEE("%s: error: failed to init llava\n", __func__);
+    auto model = llava_init(&params);
+    if (model == NULL) {
+        fprintf(stderr, "%s: error: failed to init llava model\n", __func__);
         return 1;
     }
 
-    auto image_embed = load_image(ctx_llava, &params);
-    if (!image_embed) {
-        return 1;
-    }
+    for (auto & image : params.image) {
+        auto ctx_llava = llava_init_context(&params, model);
 
-    // process the prompt
-    process_prompt(ctx_llava, image_embed, &params, params.prompt);
+        auto image_embed = load_image(ctx_llava, &params, image);
+        if (!image_embed) {
+            std::cerr << "error: failed to load image " << image << ". Terminating\n\n";
+            return 1;
+        }
+
+        // process the prompt
+        process_prompt(ctx_llava, image_embed, &params, params.prompt);
 
-    llama_print_timings(ctx_llava->ctx_llama);
+        llama_print_timings(ctx_llava->ctx_llama);
+        llava_image_embed_free(image_embed);
+        ctx_llava->model = NULL;
+        llava_free(ctx_llava);
+    }
+    llama_free_model(model);
 
-    llava_image_embed_free(image_embed);
-    llava_free(ctx_llava);
     return 0;
 }

package/src/llama.cpp/examples/lookup/lookup-stats.cpp

@@ -30,7 +30,6 @@ int main(int argc, char ** argv){
 
     // load the model
     std::tie(model, ctx) = llama_init_from_gpt_params(params);
-    llama_set_rng_seed(ctx, params.seed);
     GGML_ASSERT(llama_n_vocab(model) < (1 << 16));
 
     // tokenize the prompt

package/src/llama.cpp/examples/lookup/lookup.cpp

@@ -38,7 +38,6 @@ int main(int argc, char ** argv){
 
     // load the model
     std::tie(model, ctx) = llama_init_from_gpt_params(params);
-    llama_set_rng_seed(ctx, params.seed);
     GGML_ASSERT(llama_n_vocab(model) < (1 << 16));
 
     // tokenize the prompt

package/src/llama.cpp/examples/main/main.cpp

@@ -240,7 +240,6 @@ int main(int argc, char ** argv) {
                 return 1;
             }
             session_tokens.resize(n_token_count_out);
-            llama_set_rng_seed(ctx, params.seed);
             LOG_TEE("%s: loaded a session with prompt size of %d tokens\n", __func__, (int)session_tokens.size());
         }
     }
@@ -325,7 +324,7 @@ int main(int argc, char ** argv) {
             log_tostr(embd_inp.empty()), n_matching_session_tokens, embd_inp.size(), session_tokens.size(), embd_inp.size());
 
     // if we will use the cache for the full prompt without reaching the end of the cache, force
-    // reevaluation of the last token token to recalculate the cached logits
+    // reevaluation of the last token to recalculate the cached logits
     if (!embd_inp.empty() && n_matching_session_tokens == embd_inp.size() && session_tokens.size() > embd_inp.size()) {
         LOGLN("recalculate the cached logits (do): session_tokens.resize( %zu )", embd_inp.size() - 1);
 
@@ -363,6 +362,9 @@ int main(int argc, char ** argv) {
         params.interactive_first = true;
         params.antiprompt.emplace_back("<|im_start|>user\n");
     }
+    else if (params.conversation) {
+        params.interactive_first = true;
+    }
 
     // enable interactive mode if interactive start is specified
     if (params.interactive_first) {
@@ -545,7 +547,7 @@ int main(int argc, char ** argv) {
                 // if we run out of context:
                 // - take the n_keep first tokens from the original prompt (via n_past)
                 // - take half of the last (n_ctx - n_keep) tokens and recompute the logits in batches
-                if (n_past + (int) embd.size() + std::max<int>(0, guidance_offset) > n_ctx) {
+                if (n_past + (int) embd.size() + std::max<int>(0, guidance_offset) >= n_ctx) {
                     if (params.n_predict == -2) {
                         LOG_TEE("\n\n%s: context full and n_predict == -%d => stopping\n", __func__, params.n_predict);
                         break;
@@ -734,7 +736,7 @@ int main(int argc, char ** argv) {
         // display text
         if (input_echo && display) {
             for (auto id : embd) {
-                const std::string token_str = llama_token_to_piece(ctx, id);
+                const std::string token_str = llama_token_to_piece(ctx, id, !params.conversation);
                 printf("%s", token_str.c_str());
 
                 if (embd.size() > 1) {
@@ -797,7 +799,7 @@ int main(int argc, char ** argv) {
 
             // deal with end of generation tokens in interactive mode
             if (llama_token_is_eog(model, llama_sampling_last(ctx_sampling))) {
-                LOG("found EOS token\n");
+                LOG("found an EOG token\n");
 
                 if (params.interactive) {
                     if (!params.antiprompt.empty()) {
@@ -817,7 +819,7 @@ int main(int argc, char ** argv) {
             if (n_past > 0 && is_interacting) {
                 LOG("waiting for user input\n");
 
-                if (params.instruct || params.chatml) {
+                if (params.conversation || params.instruct || params.chatml) {
                     printf("\n> ");
                 }
 
@@ -827,7 +829,7 @@ int main(int argc, char ** argv) {
                 }
 
                 std::string buffer;
-                if (!params.input_prefix.empty()) {
+                if (!params.input_prefix.empty() && !params.conversation) {
                     LOG("appending input prefix: '%s'\n", params.input_prefix.c_str());
                     printf("%s", params.input_prefix.c_str());
                 }
@@ -851,7 +853,7 @@ int main(int argc, char ** argv) {
                 // Entering a empty line lets the user pass control back
                 if (buffer.length() > 1) {
                     // append input suffix if any
-                    if (!params.input_suffix.empty()) {
+                    if (!params.input_suffix.empty() && !params.conversation) {
                         LOG("appending input suffix: '%s'\n", params.input_suffix.c_str());
                         printf("%s", params.input_suffix.c_str());
                     }