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 14, 05:30 UTC
Update docs for Minimax-Text support (#22562)
Signed-off-by: Thomas Parnell <tpa@zurich.ibm.com>
Thomas Parnell committed
Aug 9, 2025 at 09:18 UTC
2be07a0db115e65009111145e17b034c54ae4a01
2 files changed
+6
-2
docs/models/supported_models.md
+2
-2
@@ -404,8 +404,8 @@ th {
404
| `TeleChat2ForCausalLM` | TeleChat2 | `Tele-AI/TeleChat2-3B`, `Tele-AI/TeleChat2-7B`, `Tele-AI/TeleChat2-35B`, etc. | ✅︎ | ✅︎ | ✅︎ |
405
| `TeleFLMForCausalLM` | TeleFLM | `CofeAI/FLM-2-52B-Instruct-2407`, `CofeAI/Tele-FLM`, etc. | ✅︎ | ✅︎ | ✅︎ |
406
| `XverseForCausalLM` | XVERSE | `xverse/XVERSE-7B-Chat`, `xverse/XVERSE-13B-Chat`, `xverse/XVERSE-65B-Chat`, etc. | ✅︎ | ✅︎ | ✅︎ |
407
-| `MiniMaxM1ForCausalLM` | MiniMax-Text | `MiniMaxAI/MiniMax-M1-40k`, `MiniMaxAI/MiniMax-M1-80k`, etc. | | | |
408
-| `MiniMaxText01ForCausalLM` | MiniMax-Text | `MiniMaxAI/MiniMax-Text-01`, etc. | | | |
407
+| `MiniMaxM1ForCausalLM` | MiniMax-Text | `MiniMaxAI/MiniMax-M1-40k`, `MiniMaxAI/MiniMax-M1-80k`, etc. | | | ✅︎ |
408
+| `MiniMaxText01ForCausalLM` | MiniMax-Text | `MiniMaxAI/MiniMax-Text-01`, etc. | | | ✅︎ |
409
| `Zamba2ForCausalLM` | Zamba2 | `Zyphra/Zamba2-7B-instruct`, `Zyphra/Zamba2-2.7B-instruct`, `Zyphra/Zamba2-1.2B-instruct`, etc. | | | ✅︎ |
410
411
!!! note
docs/usage/v1_guide.md
+4
@@ -111,6 +111,10 @@ Models that combine Mamba-2 and Mamba-1 layers with standard attention layers ar
111
`Zamba2ForCausalLM`, `NemotronHForCausalLM`, `FalconH1ForCausalLM` and `GraniteMoeHybridForCausalLM`, `JambaForCausalLM`). Please note that
112
these models currently require disabling prefix caching and using the FlashInfer attention backend in V1.
113
114
+Hybrid models with mechanisms different to Mamba are also supported (e.g, `MiniMaxText01ForCausalLM`, `MiniMaxM1ForCausalLM`).
115
+Please note that these models currently require disabling prefix caching, enforcing eager mode, and using the FlashInfer
116
+attention backend in V1.
117
+
118
#### Encoder-Decoder Models
119
120
Models requiring cross-attention between separate encoder and decoder (e.g., `BartForConditionalGeneration`, `MllamaForConditionalGeneration`)