Merge commit 'f0adb80bf7c2c0d80abb04f4533b5513622d9964' into concedo_experimental

# Conflicts:
#	docs/backend/CANN.md
#	docs/backend/SYCL.md
#	docs/docker.md
#	examples/sycl/run-llama2.sh
#	examples/sycl/win-run-llama2.bat
#	ggml/src/ggml-sycl/ggml-sycl.cpp
#	tools/llama-bench/README.md

Author: LostRuins

common/common.cpp

@@ -1110,6 +1110,9 @@ struct llama_model_params common_model_params_to_llama(common_params & params) {
         mparams.tensor_buft_overrides = params.tensor_buft_overrides.data();
     }
 
+    mparams.progress_callback           = params.load_progress_callback;
+    mparams.progress_callback_user_data = params.load_progress_callback_user_data;
+
     return mparams;
 }
 

common/common.h

@@ -424,6 +424,11 @@ struct common_params {
 
     // common params
     std::string out_file; // output filename for all example programs
+    // optional callback for model loading progress and cancellation:
+    // called with a progress value between 0.0 and 1.0.
+    // return false from callback to abort model loading or true to continue
+    llama_progress_callback load_progress_callback = NULL;
+    void *                  load_progress_callback_user_data = NULL;
 };
 
 // call once at the start of a program if it uses libcommon

examples/sycl/run-llama3.sh

@@ -0,0 +1,28 @@
+#!/bin/bash
+
+#  MIT license
+#  Copyright (C) 2025 Intel Corporation
+#  SPDX-License-Identifier: MIT
+
+# If you want more control, DPC++ Allows selecting a specific device through the
+# following environment variable
+#export ONEAPI_DEVICE_SELECTOR="level_zero:0"
+source /opt/intel/oneapi/setvars.sh
+
+#export GGML_SYCL_DEBUG=1
+
+#ZES_ENABLE_SYSMAN=1, Support to get free memory of GPU by sycl::aspect::ext_intel_free_memory. Recommended to use when --split-mode = layer.
+
+INPUT_PROMPT="Building a website can be done in 10 simple steps:\nStep 1:"
+MODEL_FILE=models/Meta-Llama-3.1-8B-Instruct-Q4_K_M.gguf
+NGL=99 # Layers offloaded to the GPU. If the device runs out of memory, reduce this value according to the model you are using.
+CONTEXT=4096
+
+if [ $# -gt 0 ]; then
+    GGML_SYCL_DEVICE=$1
+    echo "Using $GGML_SYCL_DEVICE as the main GPU"
+    ZES_ENABLE_SYSMAN=1 ./build/bin/llama-cli -m ${MODEL_FILE} -p "${INPUT_PROMPT}" -n 400 -e -ngl ${NGL} -c ${CONTEXT} -mg $GGML_SYCL_DEVICE -sm none
+else
+    #use multiple GPUs with same max compute units
+    ZES_ENABLE_SYSMAN=1 ./build/bin/llama-cli -m ${MODEL_FILE} -p "${INPUT_PROMPT}" -n 400 -e -ngl ${NGL} -c ${CONTEXT}
+fi

examples/sycl/win-run-llama3.bat

@@ -0,0 +1,9 @@
+::  MIT license
+::  Copyright (C) 2024 Intel Corporation
+::  SPDX-License-Identifier: MIT
+
+set INPUT2="Building a website can be done in 10 simple steps:\nStep 1:"
+@call "C:\Program Files (x86)\Intel\oneAPI\setvars.bat" intel64 --force
+
+
+.\build\bin\llama-cli.exe -m models\Meta-Llama-3.1-8B-Instruct-Q4_K_M.gguf -p %INPUT2% -n 400 -e -ngl 99

ggml/src/ggml-vulkan/ggml-vulkan.cpp

@@ -1936,18 +1936,6 @@ ModelLoadResult gpttype_load_model(const load_model_inputs inputs, FileFormat in
         clamped_max_context_length = 16384;
     }
 
-    #if defined(GGML_USE_VULKAN)
-    if (isGguf && file_format_meta.model_architecture == GGUFArch::ARCH_GLM4 && kcpp_data->n_ubatch > 16) {
-        if(debugmode==1)
-        {
-            printf("GLM-4 is broken on larger batch sizes in Vulkan. Clamp ignored in debug.\n");
-        } else {
-            printf("GLM-4 is broken on larger batch sizes in Vulkan. Clamping ubatch size to 8.\n");
-            kcpp_data->n_ubatch = 8;
-        }
-    }
-    #endif
-
     kcpp_data->n_ctx = clamped_max_context_length;
     max_context_limit_at_load = clamped_max_context_length;
     add_bos_token = !inputs.no_bos_token;

gpttype_adapter.cpp

@@ -1879,6 +1879,23 @@ Current version indicated by LITEVER below.
 	{
 		background-color: #161616;
 	}
+	.corpoendeditbutton {
+      position: fixed;
+      top: 75px;
+      right: 25px;
+	  width: 40px;
+      padding: 8px 6px;
+      background-color: #de5b5b;
+      color: white;
+      border-radius: 24px;
+	  opacity: 0.65;
+      cursor: pointer;
+      box-shadow: 1 2px 8px rgba(0, 0, 0, 0.2);
+    }
+
+    .corpoendeditbutton:hover {
+      background-color: #f67676;
+    }
 	.corpoleftpanelitems
 	{
 		display: flex;
@@ -13383,6 +13400,11 @@ Current version indicated by LITEVER below.
 		}
 		render_gametext(false);
 	}
+	function exit_editmode()
+	{
+		document.getElementById("allowediting").checked = false;
+		toggle_editable();
+	}
 
 	function toggle_hide_thinking(button) {
 		// Get the parent div of the button
@@ -18714,40 +18736,43 @@ Current version indicated by LITEVER below.
 				fulltxt = replaceAll(fulltxt, get_instruct_endtag(true), `%SpcEtg%`);
 				if (localsettings.separate_end_tags) {
 					if (get_instruct_endtag_end(true)) {
+						fulltxt = replaceAll(fulltxt, get_instruct_endtag_end(false), ``);
 						fulltxt = replaceAll(fulltxt, get_instruct_endtag_end(true), ``);
 					}
 					if (get_instruct_starttag_end(true)) {
+						fulltxt = replaceAll(fulltxt, get_instruct_starttag_end(false), ``);
 						fulltxt = replaceAll(fulltxt, get_instruct_starttag_end(true), ``);
 					}
 				}
 
-				if(localsettings.instruct_has_markdown && synchro_pending_stream=="")
-				{
-					//if a list has a starttag on the same line, add a newline before it
-					fulltxt = fulltxt.replace(/(\n[-*] .+?)(%SpcStg%)/g, "$1\n$2");
-					let codeblockcount = (fulltxt.match(/```/g) || []).length;
-					if(codeblockcount>0 && codeblockcount%2!=0 )
-					{
-						fulltxt += "```"; //force end code block
-					}
-					fulltxt = simpleMarkdown(fulltxt,localsettings.instruct_has_latex);
-				}
-
 				let instruct_turns = repack_instruct_turns(fulltxt, `%SpcStg%`,`%SpcEtg%`, true);
 				fulltxt = "";
 				for(let i=0;i<instruct_turns.length;++i)
 				{
 					let curr = instruct_turns[i];
+					let currmsg = curr.msg;
+					if(localsettings.instruct_has_markdown && (synchro_pending_stream==""||(i+1)<instruct_turns.length))
+					{
+						//if a list has a starttag on the same line, add a newline before it
+						currmsg = currmsg.replace(/(\n[-*] .+?)(%SpcStg%)/g, "$1\n$2");
+						let codeblockcount = (currmsg.match(/```/g) || []).length;
+						if(codeblockcount>0 && codeblockcount%2!=0 )
+						{
+							currmsg += "```"; //force end code block
+						}
+						currmsg = simpleMarkdown(currmsg,localsettings.instruct_has_latex);
+					}
+
 					if(curr.myturn)
 					{
-						fulltxt += `<hr style="margin-top: 12px; margin-bottom: 12px;"><span class="color_cyan"><img src="${human_square}" style="height:38px;width:auto;padding:3px 6px 3px 3px;border-radius: 8%;"/>${curr.msg}</span>`;
+						fulltxt += `<hr style="margin-top: 12px; margin-bottom: 12px;"><span class="color_cyan"><img src="${human_square}" style="height:38px;width:auto;padding:3px 6px 3px 3px;border-radius: 8%;"/>${currmsg}</span>`;
 					}
 					else
 					{
 						if (i == 0) {
-							fulltxt += `${curr.msg}`;
+							fulltxt += `${currmsg}`;
 						} else {
-							fulltxt += `<hr style="margin-top: 12px; margin-bottom: 12px;"><img src="${niko_square}" style="height:38px;width:auto;padding:3px 6px 3px 3px;border-radius: 8%;"/>${curr.msg}</span>`;
+							fulltxt += `<hr style="margin-top: 12px; margin-bottom: 12px;"><img src="${niko_square}" style="height:38px;width:auto;padding:3px 6px 3px 3px;border-radius: 8%;"/>${currmsg}</span>`;
 						}
 					}
 				}
@@ -19070,6 +19095,12 @@ Current version indicated by LITEVER below.
 			document.getElementById("normalinterface").classList.remove("hidden");
 		}
 
+		if (inEditMode && isCorpoUiStyle) {
+			document.getElementById("corpoendedit").classList.remove("hidden");
+		} else {
+			document.getElementById("corpoendedit").classList.add("hidden");
+		}
+
 		update_submit_button(true); //full update for submit button, otherwise just text when not generating
 
 		document.getElementById("btnautogenmem").disabled = document.getElementById("btnsend").disabled;
@@ -19803,7 +19834,7 @@ Current version indicated by LITEVER below.
 
 	function show_corpo_leftpanel(open)
 	{
-		let panel = document.getElementById('corpo_leftpannel');
+		let panel = document.getElementById('corpo_leftpanel');
 		if(open)
 		{
 			panel.classList.add('open');
@@ -21945,9 +21976,10 @@ Current version indicated by LITEVER below.
 			</div>
 		</div>
 
+		<button title="Exit Edit Mode" id="corpoendedit" class="corpoendeditbutton hidden" onclick="exit_editmode()">X</button>
 		<div id="corpointerface" class="hidden">
 			<div class="corpostyle">
-				<div id="corpo_leftpannel" class="corpoleftpanel">
+				<div id="corpo_leftpanel" class="corpoleftpanel">
 					<button title="Hide Corpo Side Panel" class="corpo_leftpanel_close mainnav" onclick="show_corpo_leftpanel(false)">&times;</button>
 					<div id="corpoleftpanelitems" class="corpoleftpanelitems">
 					<div class="corpoleftpanelitemstopper" id="corpoleftpanelitemstopper"></div>