mirror of
https://github.com/ggml-org/llama.cpp.git
synced 2025-11-06 09:46:50 +00:00
cd5e3b5754
* server : support unified context across slots * cont : fix speculative decoding initialization * context : fix n_ctx_per_seq computation * server : purge slots one by one * tests : add unified cache server tests * llama : update per-seq context computation * test-thread-safety : handle tiny training context of the input model * server : fix server_tokens clear() * server : use 4 slots + unified KV by default * llama : add note about context size queries * cont : update todos [no ci] * context : do not cap the size of the context * tests : adjust parameters to be CI friendlier * context : add warning
78 lines
2.5 KiB
Python
78 lines
2.5 KiB
Python
import pytest
|
|
from utils import *
|
|
|
|
server = ServerPreset.tinyllama_infill()
|
|
|
|
@pytest.fixture(autouse=True)
|
|
def create_server():
|
|
global server
|
|
server = ServerPreset.tinyllama_infill()
|
|
|
|
|
|
def test_infill_without_input_extra():
|
|
global server
|
|
server.start()
|
|
res = server.make_request("POST", "/infill", data={
|
|
"input_prefix": "#include <cstdio>\n#include \"llama.h\"\n\nint main() {\n",
|
|
"prompt": " int n_threads = llama_",
|
|
"input_suffix": "}\n",
|
|
})
|
|
assert res.status_code == 200
|
|
assert match_regex("(Ann|small|shiny|Daddy|Jimmy)+", res.body["content"])
|
|
|
|
|
|
def test_infill_with_input_extra():
|
|
global server
|
|
server.start()
|
|
res = server.make_request("POST", "/infill", data={
|
|
"input_extra": [{
|
|
"filename": "llama.h",
|
|
"text": "LLAMA_API int32_t llama_n_threads();\n"
|
|
}],
|
|
"input_prefix": "#include <cstdio>\n#include \"llama.h\"\n\nint main() {\n",
|
|
"prompt": " int n_threads = llama_",
|
|
"input_suffix": "}\n",
|
|
})
|
|
assert res.status_code == 200
|
|
assert match_regex("(Dad|excited|park|Jimmy)+", res.body["content"])
|
|
|
|
|
|
@pytest.mark.parametrize("input_extra", [
|
|
{},
|
|
{"filename": "ok"},
|
|
{"filename": 123},
|
|
{"filename": 123, "text": "abc"},
|
|
{"filename": 123, "text": 456},
|
|
])
|
|
def test_invalid_input_extra_req(input_extra):
|
|
global server
|
|
server.start()
|
|
res = server.make_request("POST", "/infill", data={
|
|
"input_extra": [input_extra],
|
|
"input_prefix": "#include <cstdio>\n#include \"llama.h\"\n\nint main() {\n",
|
|
"prompt": " int n_threads = llama_",
|
|
"input_suffix": "}\n",
|
|
})
|
|
assert res.status_code == 400
|
|
assert "error" in res.body
|
|
|
|
|
|
@pytest.mark.skipif(not is_slow_test_allowed(), reason="skipping slow test")
|
|
def test_with_qwen_model():
|
|
global server
|
|
server.model_file = None
|
|
server.model_hf_repo = "ggml-org/Qwen2.5-Coder-1.5B-IQ3_XXS-GGUF"
|
|
server.model_hf_file = "qwen2.5-coder-1.5b-iq3_xxs-imat.gguf"
|
|
server.start(timeout_seconds=600)
|
|
res = server.make_request("POST", "/infill", data={
|
|
"input_extra": [{
|
|
"filename": "llama.h",
|
|
"text": "LLAMA_API int32_t llama_n_threads();\n"
|
|
}],
|
|
"input_prefix": "#include <cstdio>\n#include \"llama.h\"\n\nint main() {\n",
|
|
"prompt": " int n_threads = llama_",
|
|
"input_suffix": "}\n",
|
|
})
|
|
assert res.status_code == 200
|
|
assert res.body["content"] == "n_threads();\n printf(\"Number of threads: %d\\n\", n_threads);\n return 0;\n"
|