biondizzle/vllm
1
0
Fork 0
Code Issues Pull Requests Actions 2 Packages Projects Releases Wiki Activity

[Bugfix] Fix PP for ChatGLM and Molmo (#9422)

Browse Source
This commit is contained in:
Cyrus Leung
2024-10-24 14:12:05 +08:00
committed by GitHub
parent 056a68c7db
commit 836e8ef6ee
7 changed files with 195 additions and 122 deletions
Download Patch File Download Diff File Expand all files Collapse all files

37
tests/distributed/test_pipeline_parallel.py
View File

@@ -118,11 +118,8 @@ class PPTestSettings:
# The values displayed here are only a rough indicator of the size of the model
# yapf: disable
GENERATION_MODEL_SETTINGS = {
# [DETAILED TESTS]
"meta-llama/Meta-Llama-3-8B": PPTestSettings.detailed(),
"microsoft/Phi-3-mini-4k-instruct": PPTestSettings.detailed(trust_remote_code=True, multi_node_only=True), # noqa: E501
# [FAST TESTS]
TEXT_GENERATION_MODELS = {
# [Decoder-only]
# Uses Llama
# "BAAI/AquilaChat-7B": PPTestSettings.fast(),
"Snowflake/snowflake-arctic-instruct": PPTestSettings.fast(tp_base=8, trust_remote_code=True), # noqa: E501
@@ -151,6 +148,7 @@ GENERATION_MODEL_SETTINGS = {
"core42/jais-13b-chat": PPTestSettings.fast(),
# TODO: Implement PP
# "ai21labs/AI21-Jamba-1.5-Mini": PPTestSettings.fast(),
"meta-llama/Meta-Llama-3-8B": PPTestSettings.detailed(),
"openbmb/MiniCPM-2B-sft-bf16": PPTestSettings.fast(trust_remote_code=True),
"openbmb/MiniCPM3-4B": PPTestSettings.fast(trust_remote_code=True),
# Uses Llama
@@ -163,6 +161,7 @@ GENERATION_MODEL_SETTINGS = {
"facebook/opt-iml-max-1.3b": PPTestSettings.fast(),
"OrionStarAI/Orion-14B-Chat": PPTestSettings.fast(trust_remote_code=True),
"microsoft/phi-2": PPTestSettings.fast(),
"microsoft/Phi-3-mini-4k-instruct": PPTestSettings.detailed(trust_remote_code=True, multi_node_only=True), # noqa: E501
"microsoft/Phi-3-small-8k-instruct": PPTestSettings.fast(trust_remote_code=True), # noqa: E501
"microsoft/Phi-3.5-MoE-instruct": PPTestSettings.fast(trust_remote_code=True), # noqa: E501
"adept/persimmon-8b-chat": PPTestSettings.fast(),
@@ -174,40 +173,40 @@ GENERATION_MODEL_SETTINGS = {
"upstage/solar-pro-preview-instruct": PPTestSettings.fast(tp_base=2),
# FIXME: Cannot load tokenizer in latest transformers version
# "xverse/XVERSE-7B-Chat": PPTestSettings.fast(trust_remote_code=True),
# [Encoder-only]
# TODO: Implement PP
# "facebook/bart-base": PPTestSettings.fast(),
}
EMBEDDING_MODEL_SETTINGS = { # type: ignore[var-annotated]
# [FAST TESTS]
EMBEDDING_MODELS = { # type: ignore[var-annotated]
# [Text-only]
"intfloat/e5-mistral-7b-instruct": PPTestSettings.fast(),
"BAAI/bge-multilingual-gemma2": PPTestSettings.fast(),
"Qwen/Qwen2.5-Math-RM-72B": PPTestSettings.fast(tp_base=4, trust_remote_code=True), # noqa: E501
}
MULTIMODAL_MODEL_SETTINGS = {
# [FAST TESTS]
MULTIMODAL_MODELS = {
# [Decoder-only]
"Salesforce/blip2-opt-2.7b": PPTestSettings.fast(),
"facebook/chameleon-7b": PPTestSettings.fast(),
"adept/fuyu-8b": PPTestSettings.fast(),
"THUDM/glm-4v-9b": PPTestSettings.fast(trust_remote_code=True),
"OpenGVLab/InternVL2-1B": PPTestSettings.fast(trust_remote_code=True),
"llava-hf/llava-1.5-7b-hf": PPTestSettings.fast(),
"llava-hf/llava-v1.6-mistral-7b-hf": PPTestSettings.fast(),
"llava-hf/LLaVA-NeXT-Video-7B-hf": PPTestSettings.fast(),
"llava-hf/llava-onevision-qwen2-0.5b-ov-hf": PPTestSettings.fast(),
"openbmb/MiniCPM-Llama3-V-2_5": PPTestSettings.fast(trust_remote_code=True),
# TODO: Implement PP
# "meta-llama/Llama-3.2-11B-Vision-Instruct": PPTestSettings.fast(),
"allenai/Molmo-7B-D-0924": PPTestSettings.fast(trust_remote_code=True),
"microsoft/Phi-3-vision-128k-instruct": PPTestSettings.fast(trust_remote_code=True), # noqa: E501
"mistralai/Pixtral-12B-2409": PPTestSettings.fast(tp_base=2, tokenizer_mode="mistral"), # noqa: E501
"Qwen/Qwen-VL-Chat": PPTestSettings.fast(trust_remote_code=True),
"Qwen/Qwen2-Audio-7B-Instruct": PPTestSettings.fast(),
"Qwen/Qwen2-VL-2B-Instruct": PPTestSettings.fast(),
"fixie-ai/ultravox-v0_3": PPTestSettings.fast(),
}
CONDITIONAL_GENERATION_MODEL_SETTINGS = { # type: ignore[var-annotated]
# [FAST TESTS]
# [Encoder-decoder]
# TODO: Implement PP
# "facebook/bart-base": PPTestSettings.fast(),
# "meta-llama/Llama-3.2-11B-Vision-Instruct": PPTestSettings.fast(),
}
# yapf: enable
@@ -323,7 +322,7 @@ def _compare_tp(
("model_name", "parallel_setup", "distributed_backend", "task",
"test_options"),
[
params for model_name, settings in GENERATION_MODEL_SETTINGS.items()
params for model_name, settings in TEXT_GENERATION_MODELS.items()
for params in settings.iter_params(model_name)
if model_name in TEST_MODELS
],
@@ -350,7 +349,7 @@ def test_tp_language_generation(
("model_name", "parallel_setup", "distributed_backend", "task",
"test_options"),
[
params for model_name, settings in EMBEDDING_MODEL_SETTINGS.items()
params for model_name, settings in EMBEDDING_MODELS.items()
for params in settings.iter_params(model_name)
if model_name in TEST_MODELS
],
@@ -377,7 +376,7 @@ def test_tp_language_embedding(
("model_name", "parallel_setup", "distributed_backend", "task",
"test_options"),
[
params for model_name, settings in MULTIMODAL_MODEL_SETTINGS.items()
params for model_name, settings in MULTIMODAL_MODELS.items()
for params in settings.iter_params(model_name)
if model_name in TEST_MODELS
],