@tduncan / Nvidia-Optimization-Stack-VLLM Mirror · read-only / commits / 4f341bd4bf
Fork 0

A high-throughput and memory-efficient inference and serving engine for LLMs

Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git

Synced Nov 16, 10:45 UTC

[Doc] Update list of supported models (#8987)

Cyrus Leung committed Oct 2, 2024 at 00:35 UTC 4f341bd4bf35c5b431dc523bab86e4ae210baaf8
2 files changed +21 -8
docs/source/models/supported_models.rst
+16 -4
@@ -20,7 +20,7 @@ Decoder-only Language Models
20 - Example HuggingFace Models
21 - :ref:`LoRA <lora>`
22 * - :code:`AquilaForCausalLM`
23 - - Aquila & Aquila2
23 + - Aquila, Aquila2
24 - :code:`BAAI/Aquila-7B`, :code:`BAAI/AquilaChat-7B`, etc.
25 - ✅︎
26 * - :code:`ArcticForCausalLM`
@@ -28,7 +28,7 @@ Decoder-only Language Models
28 - :code:`Snowflake/snowflake-arctic-base`, :code:`Snowflake/snowflake-arctic-instruct`, etc.
29 -
30 * - :code:`BaiChuanForCausalLM`
31 - - Baichuan & Baichuan2
31 + - Baichuan2, Baichuan
32 - :code:`baichuan-inc/Baichuan2-13B-Chat`, :code:`baichuan-inc/Baichuan-7B`, etc.
33 - ✅︎
34 * - :code:`BloomForCausalLM`
@@ -51,6 +51,14 @@ Decoder-only Language Models
51 - DeciLM
52 - :code:`Deci/DeciLM-7B`, :code:`Deci/DeciLM-7B-instruct`, etc.
53 -
54 + * - :code:`DeepseekForCausalLM`
55 + - DeepSeek
56 + - :code:`deepseek-ai/deepseek-llm-67b-base`, :code:`deepseek-ai/deepseek-llm-7b-chat` etc.
57 + -
58 + * - :code:`DeepseekV2ForCausalLM`
59 + - DeepSeek-V2
60 + - :code:`deepseek-ai/DeepSeek-V2`, :code:`deepseek-ai/DeepSeek-V2-Chat` etc.
61 + -
62 * - :code:`ExaoneForCausalLM`
63 - EXAONE-3
64 - :code:`LGAI-EXAONE/EXAONE-3.0-7.8B-Instruct`, etc.
@@ -83,6 +91,10 @@ Decoder-only Language Models
91 - GPT-NeoX, Pythia, OpenAssistant, Dolly V2, StableLM
92 - :code:`EleutherAI/gpt-neox-20b`, :code:`EleutherAI/pythia-12b`, :code:`OpenAssistant/oasst-sft-4-pythia-12b-epoch-3.5`, :code:`databricks/dolly-v2-12b`, :code:`stabilityai/stablelm-tuned-alpha-7b`, etc.
93 -
94 + * - :code:`GraniteForCausalLM`
95 + - Granite, Power-LM
96 + - :code:`ibm/granite-7b-base`, :code:`ibm/PowerLM-3b` etc.
97 + - ✅︎
98 * - :code:`InternLMForCausalLM`
99 - InternLM
100 - :code:`internlm/internlm-7b`, :code:`internlm/internlm-chat-7b`, etc.
@@ -97,7 +109,7 @@ Decoder-only Language Models
109 -
110 * - :code:`JambaForCausalLM`
111 - Jamba
100 - - :code:`ai21labs/Jamba-v0.1`, etc.
112 + - :code:`ai21labs/AI21-Jamba-1.5-Large`, :code:`ai21labs/AI21-Jamba-1.5-Mini`, :code:`ai21labs/Jamba-v0.1`, etc.
113 - ✅︎
114 * - :code:`LlamaForCausalLM`
115 - Llama 3.1, Llama 3, Llama 2, LLaMA, Yi
@@ -177,7 +189,7 @@ Decoder-only Language Models
189 -
190 * - :code:`StableLmForCausalLM`
191 - StableLM
180 - - :code:`stabilityai/stablelm-3b-4e1t/` , :code:`stabilityai/stablelm-base-alpha-7b-v2`, etc.
192 + - :code:`stabilityai/stablelm-3b-4e1t`, :code:`stabilityai/stablelm-base-alpha-7b-v2`, etc.
193 -
194 * - :code:`Starcoder2ForCausalLM`
195 - Starcoder2
vllm/model_executor/models/__init__.py
+5 -4
@@ -12,6 +12,7 @@ logger = init_logger(__name__)
12 _GENERATION_MODELS = {
13 "AquilaModel": ("llama", "LlamaForCausalLM"),
14 "AquilaForCausalLM": ("llama", "LlamaForCausalLM"), # AquilaChat2
15 + "ArcticForCausalLM": ("arctic", "ArcticForCausalLM"),
16 "BaiChuanForCausalLM": ("baichuan", "BaiChuanForCausalLM"), # baichuan-7b
17 "BaichuanForCausalLM": ("baichuan", "BaichuanForCausalLM"), # baichuan-13b
18 "BloomForCausalLM": ("bloom", "BloomForCausalLM"),
@@ -30,9 +31,11 @@ _GENERATION_MODELS = {
31 "GPTBigCodeForCausalLM": ("gpt_bigcode", "GPTBigCodeForCausalLM"),
32 "GPTJForCausalLM": ("gpt_j", "GPTJForCausalLM"),
33 "GPTNeoXForCausalLM": ("gpt_neox", "GPTNeoXForCausalLM"),
34 + "GraniteForCausalLM": ("granite", "GraniteForCausalLM"),
35 "InternLMForCausalLM": ("llama", "LlamaForCausalLM"),
36 "InternLM2ForCausalLM": ("internlm2", "InternLM2ForCausalLM"),
37 "JAISLMHeadModel": ("jais", "JAISLMHeadModel"),
38 + "JambaForCausalLM": ("jamba", "JambaForCausalLM"),
39 "LlamaForCausalLM": ("llama", "LlamaForCausalLM"),
40 # For decapoda-research/llama-*
41 "LLaMAForCausalLM": ("llama", "LlamaForCausalLM"),
@@ -52,6 +55,7 @@ _GENERATION_MODELS = {
55 "PersimmonForCausalLM": ("persimmon", "PersimmonForCausalLM"),
56 "PhiForCausalLM": ("phi", "PhiForCausalLM"),
57 "Phi3ForCausalLM": ("phi3", "Phi3ForCausalLM"),
58 + "Phi3SmallForCausalLM": ("phi3_small", "Phi3SmallForCausalLM"),
59 "PhiMoEForCausalLM": ("phimoe", "PhiMoEForCausalLM"),
60 "Qwen2ForCausalLM": ("qwen2", "Qwen2ForCausalLM"),
61 "Qwen2MoeForCausalLM": ("qwen2_moe", "Qwen2MoeForCausalLM"),
@@ -62,14 +66,11 @@ _GENERATION_MODELS = {
66 "StableLmForCausalLM": ("stablelm", "StablelmForCausalLM"),
67 "Starcoder2ForCausalLM": ("starcoder2", "Starcoder2ForCausalLM"),
68 "SolarForCausalLM": ("solar", "SolarForCausalLM"),
65 - "ArcticForCausalLM": ("arctic", "ArcticForCausalLM"),
69 "XverseForCausalLM": ("xverse", "XverseForCausalLM"),
67 - "Phi3SmallForCausalLM": ("phi3_small", "Phi3SmallForCausalLM"),
70 + # NOTE: The below models are for speculative decoding only
71 "MedusaModel": ("medusa", "Medusa"),
72 "EAGLEModel": ("eagle", "EAGLE"),
73 "MLPSpeculatorPreTrainedModel": ("mlp_speculator", "MLPSpeculator"),
71 - "JambaForCausalLM": ("jamba", "JambaForCausalLM"),
72 - "GraniteForCausalLM": ("granite", "GraniteForCausalLM")
74 }
75
76 _EMBEDDING_MODELS = {