Merge branch 'upstream' into concedo_experimental

# Conflicts:
#	.devops/openvino.Dockerfile
#	.github/workflows/build-cache.yml
#	.github/workflows/build-openvino.yml
#	.github/workflows/build-self-hosted.yml
#	.github/workflows/release.yml
#	ci/run.sh
#	docs/backend/OPENVINO.md
#	docs/speculative.md
#	ggml/src/ggml-hexagon/ggml-hexagon.cpp
#	ggml/src/ggml-hexagon/htp/htp-ops.h
#	ggml/src/ggml-hexagon/htp/hvx-arith.h
#	ggml/src/ggml-hexagon/htp/hvx-log.h
#	ggml/src/ggml-hexagon/htp/main.c
#	ggml/src/ggml-hexagon/htp/unary-ops.c
#	ggml/src/ggml-hexagon/htp/unary-ops.h
#	ggml/src/ggml-opencl/ggml-opencl.cpp
#	ggml/src/ggml-openvino/CMakeLists.txt
#	ggml/src/ggml-openvino/ggml-decoder.cpp
#	ggml/src/ggml-openvino/ggml-decoder.h
#	ggml/src/ggml-openvino/ggml-openvino-extra.cpp
#	ggml/src/ggml-openvino/ggml-openvino.cpp
#	ggml/src/ggml-openvino/openvino/op/cpy.cpp
#	ggml/src/ggml-openvino/openvino/op/flash_attn_ext.cpp
#	ggml/src/ggml-openvino/openvino/op/gated_delta_net.cpp
#	ggml/src/ggml-openvino/openvino/op/view.cpp
#	ggml/src/ggml-openvino/openvino/op_table.cpp
#	ggml/src/ggml-openvino/openvino/op_table.h
#	ggml/src/ggml-openvino/openvino/translate_session.cpp
#	ggml/src/ggml-openvino/openvino/utils.cpp
#	ggml/src/ggml-openvino/utils.cpp
#	ggml/src/ggml-openvino/utils.h
#	ggml/src/ggml-sycl/fattn-onednn.cpp
#	ggml/src/ggml-sycl/fattn.cpp
#	scripts/pr2wt.sh
#	src/CMakeLists.txt
#	src/llama-mmap.cpp
#	src/llama-quant.cpp
#	tests/CMakeLists.txt
#	tests/test-arg-parser.cpp
#	tests/test-backend-ops.cpp
#	tests/test-llama-archs.cpp
#	tests/test-save-load-state.cpp
#	tools/cli/README.md
#	tools/completion/README.md
#	tools/server/README.md

tools/server/tests/unit/test_speculative.py

@@ -52,6 +52,18 @@ def test_with_and_without_draft():
 
     assert tokens_no_draft == tokens_draft
 
+    server.stop()
+    create_server()
+    assert server.spec_draft_n_max is not None
+    server.spec_synth_rates = [0.0] * server.spec_draft_n_max
+    server.start()
+    res = server.make_request("POST", "/completion", data=request)
+
+    assert res.status_code == 200
+    assert res.body["timings"]["draft_n"] > 0
+    assert res.body["timings"]["draft_n_accepted"] == 0
+    assert res.body["tokens"] == tokens_no_draft
+
 
 def test_different_draft_min_draft_max():
     global server
@@ -80,6 +92,66 @@ def test_different_draft_min_draft_max():
         last_content = res.body["content"]
 
 
+def test_synth_is_deterministic():
+    global server
+    assert server.spec_draft_n_max is not None
+    server.spec_synth_rates = [0.75 ** (i + 1) for i in range(server.spec_draft_n_max)]
+    server.start()
+
+    request = {
+        "prompt": "I believe the meaning of life is",
+        "temperature": 0.2,
+        "top_k": 5,
+        "seed": 4242,
+        "n_predict": 32,
+    }
+    responses = [server.make_request("POST", "/completion", data=request) for _ in range(2)]
+
+    for res in responses:
+        assert res.status_code == 200
+        assert res.body["timings"]["draft_n"] > 0
+    assert responses[0].body["timings"]["draft_n"] == responses[1].body["timings"]["draft_n"]
+    assert responses[0].body["timings"]["draft_n_accepted"] == responses[1].body["timings"]["draft_n_accepted"]
+
+
+def test_synth_ignores_target_tokens():
+    global server
+    assert server.spec_draft_n_max is not None
+    server.spec_synth_rates = [1.0] * server.spec_draft_n_max
+    server.start()
+
+    res = server.make_request("POST", "/completion", data={
+        "prompt": "I believe the meaning of life is",
+        "temperature": 0.0,
+        "seed": 4242,
+        "n_predict": 32,
+    })
+
+    assert res.status_code == 200
+    assert res.body["timings"]["draft_n"] > 0
+    assert res.body["timings"]["draft_n_accepted"] == res.body["timings"]["draft_n"]
+
+    res = server.make_request("POST", "/completion", data={
+        "prompt": "I believe the meaning of life is",
+        "temperature": 0.0,
+        "seed": 4242,
+        "n_predict": 6,
+        "grammar": 'root ::= "a"{5,5}',
+    })
+    assert res.status_code == 200, res.body
+
+    res = server.make_request("POST", "/completion", data={
+        "prompt": "Respond with only: OK",
+        "temperature": 0.0,
+        "seed": 4242,
+        "n_predict": 64,
+        "ignore_eos": True,
+    })
+    assert res.status_code == 200, res.body
+    assert res.body["tokens_predicted"] == 64
+    assert res.body["stop_type"] == "limit"
+
+
 def test_slot_ctx_not_exceeded():
     global server
     server.n_ctx = 256

tools/server/tests/utils.py

@@ -99,6 +99,8 @@ class ServerProcess:
     spec_type: str | None = None
     spec_draft_n_min: int | None = None
     spec_draft_n_max: int | None = None
+    spec_synth_len: float | None = None
+    spec_synth_rates: List[float] | None = None
     no_ui: bool | None = None
     jinja: bool | None = None
     reasoning_format: Literal['deepseek', 'none', 'nothink'] | None = None
@@ -245,6 +247,11 @@ class ServerProcess:
             server_args.extend(["--spec-draft-n-max", self.spec_draft_n_max])
         if self.spec_draft_n_min:
             server_args.extend(["--spec-draft-n-min", self.spec_draft_n_min])
+        if self.spec_synth_len is not None:
+            server_args.extend(["--spec-synth-len", self.spec_synth_len])
+        if self.spec_synth_rates is not None:
+            rates = ",".join(str(rate) for rate in self.spec_synth_rates)
+            server_args.extend(["--spec-synth-rates", rates])
         if self.no_ui:
             server_args.append("--no-ui")
         if self.no_models_autoload: