model: support teleflm (#10573)

docs/supported_models/generative_models.md

@@ -52,6 +52,11 @@ in the GitHub search bar.
 | **Arcee AFM-4.5B**               | `arcee-ai/AFM-4.5B-Base`                         | Arcee's foundational model series for real world reliability and edge deployments. |
 | **Persimmon** (8B)               | `adept/persimmon-8b-chat`                         | Adept’s open 8B model with a 16K context window and fast inference; trained for broad usability and licensed under Apache 2.0. |
 | **Solar** (10.7B)               | `upstage/SOLAR-10.7B-Instruct-v1.0`                         | Upstage's 10.7B parameter model, optimized for instruction-following tasks. This architecture incorporates a depth-up scaling methodology, enhancing model performance. |
+
+
+| **Tele FLM** (52B-1T)               | `CofeAI/Tele-FLM`                         | BAAI & TeleAI's multilingual model, available in 52-billion and 1-trillion parameter variants. It is a decoder-only transformer trained on ~2T tokens |
+
+
 | **Ling** (16.8B–290B) | `inclusionAI/Ling-lite`, `inclusionAI/Ling-plus` | InclusionAI’s open MoE models. Ling-Lite has 16.8B total / 2.75B active parameters, and Ling-Plus has 290B total / 28.8B active parameters. They are designed for high performance on NLP and complex reasoning tasks. |
 | **Granite 3.0, 3.1** (IBM)               | `ibm-granite/granite-3.1-8b-instruct`                          | IBM's open dense foundation models optimized for reasoning, code, and business AI use cases. Integrated with Red Hat and watsonx systems. |
 | **Granite 3.0 MoE** (IBM)               | `ibm-granite/granite-3.0-3b-a800m-instruct`                          | IBM’s Mixture-of-Experts models offering strong performance with cost-efficiency. MoE expert routing designed for enterprise deployment at scale. |

python/sglang/srt/models/teleflm.py

@@ -0,0 +1,104 @@
+# SPDX-License-Identifier: Apache-2.0
+# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
+
+# Adapted from
+# https://github.com/huggingface/transformers/blob/v4.28.0/src/transformers/models/llama/modeling_llama.py
+# Copyright 2023 The vLLM team.
+# Copyright 2022 EleutherAI and the HuggingFace Inc. team. All rights reserved.
+#
+# This code is based on EleutherAI's GPT-NeoX library and the GPT-NeoX
+# and OPT implementations in this library. It has been modified from its
+# original forms to accommodate minor architectural differences compared
+# to GPT-NeoX and OPT used by the Meta AI team that trained the model.
+#
+# Licensed under the Apache License, Version 2.0 (the "License");
+# you may not use this file except in compliance with the License.
+# You may obtain a copy of the License at
+#
+#     http://www.apache.org/licenses/LICENSE-2.0
+#
+# Unless required by applicable law or agreed to in writing, software
+# distributed under the License is distributed on an "AS IS" BASIS,
+# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+# See the License for the specific language governing permissions and
+# limitations under the License.
+# Adapted from https://github.com/vllm-project/vllm/blob/main/vllm/model_executor/models/teleflm.py
+
+from typing import List, Optional, Tuple, Union
+
+import torch
+from transformers import LlamaConfig
+
+from sglang.srt.layers.quantization.base_config import QuantizationConfig
+from sglang.srt.model_executor.forward_batch_info import ForwardBatch, PPProxyTensors
+from sglang.srt.models.llama import LlamaForCausalLM, LlamaModel
+
+
+class TeleFLMModel(LlamaModel):
+    """
+    This implementation is based on the µScaling paper presented at
+    the ICLR 2025 Workshop:
+    NanoLM: An Affordable LLM Study Benchmark \
+    via Accurate Loss Prediction across Scales
+    by Yiqun Yao et al.
+    Available at: https://openreview.net/forum?id=IwaPYg1SCA
+    arXiv preprint: https://arxiv.org/abs/2304.06875
+    """
+
+    def __init__(
+        self,
+        config: LlamaConfig,
+        quant_config: Optional[QuantizationConfig] = None,
+        prefix: str = "",
+    ) -> None:
+        super().__init__(config, quant_config=quant_config, prefix=prefix)
+        self.use_mup = getattr(self.config, "use_mup", False)
+        if self.use_mup:
+            self.input_mult = self.config.input_mult
+
+    def forward(
+        self,
+        input_ids: torch.Tensor,
+        positions: torch.Tensor,
+        forward_batch: ForwardBatch,
+        input_embeds: torch.Tensor = None,
+        pp_proxy_tensors: Optional[PPProxyTensors] = None,
+    ) -> Union[torch.Tensor, Tuple[torch.Tensor, List[torch.Tensor]], PPProxyTensors]:
+        if self.pp_group.is_first_rank and input_embeds is None:
+            input_embeds = self.embed_tokens(input_ids)
+            if self.use_mup:
+                input_embeds = input_embeds * self.input_mult
+
+        return super().forward(
+            input_ids=input_ids,
+            positions=positions,
+            forward_batch=forward_batch,
+            input_embeds=input_embeds,
+            pp_proxy_tensors=pp_proxy_tensors,
+        )
+
+
+class TeleFLMForCausalLM(LlamaForCausalLM):
+    def __init__(
+        self,
+        config: LlamaConfig,
+        quant_config: Optional[QuantizationConfig] = None,
+        prefix: str = "",
+    ):
+        super().__init__(config, quant_config=quant_config, prefix=prefix)
+        self.use_mup = getattr(self.config, "use_mup", False)
+        if self.use_mup:
+            self.mup_scale_factor = self.config.mup_scale_factor
+            self.output_mult = self.config.output_mult / self.mup_scale_factor
+            self.logits_processor.logit_scale = self.output_mult
+
+    def _init_model(
+        self,
+        config: LlamaConfig,
+        quant_config: Optional[QuantizationConfig] = None,
+        prefix: str = "",
+    ):
+        return TeleFLMModel(config, quant_config=quant_config, prefix=prefix)
+
+
+EntryClass = TeleFLMForCausalLM