implement mamba

Dockerfile

@@ -13,7 +13,7 @@ ARG TARGETVARIANT
 
 ENV BUILD_TYPE=${BUILD_TYPE}
 
-ENV EXTERNAL_GRPC_BACKENDS="coqui:/build/backend/python/coqui/run.sh,huggingface-embeddings:/build/backend/python/sentencetransformers/run.sh,petals:/build/backend/python/petals/run.sh,transformers:/build/backend/python/transformers/run.sh,sentencetransformers:/build/backend/python/sentencetransformers/run.sh,autogptq:/build/backend/python/autogptq/run.sh,bark:/build/backend/python/bark/run.sh,diffusers:/build/backend/python/diffusers/run.sh,exllama:/build/backend/python/exllama/run.sh,vall-e-x:/build/backend/python/vall-e-x/run.sh,vllm:/build/backend/python/vllm/run.sh,exllama2:/build/backend/python/exllama2/run.sh,transformers-musicgen:/build/backend/python/transformers-musicgen/run.sh"
+ENV EXTERNAL_GRPC_BACKENDS="coqui:/build/backend/python/coqui/run.sh,huggingface-embeddings:/build/backend/python/sentencetransformers/run.sh,petals:/build/backend/python/petals/run.sh,transformers:/build/backend/python/transformers/run.sh,sentencetransformers:/build/backend/python/sentencetransformers/run.sh,autogptq:/build/backend/python/autogptq/run.sh,bark:/build/backend/python/bark/run.sh,diffusers:/build/backend/python/diffusers/run.sh,exllama:/build/backend/python/exllama/run.sh,vall-e-x:/build/backend/python/vall-e-x/run.sh,vllm:/build/backend/python/vllm/run.sh,mamba:/build/backend/python/mamba/run.sh,exllama2:/build/backend/python/exllama2/run.sh,transformers-musicgen:/build/backend/python/transformers-musicgen/run.sh"
 
 ARG GO_TAGS="stablediffusion tinydream tts"
 
@@ -168,6 +168,9 @@ RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
 RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
 	PATH=$PATH:/opt/conda/bin make -C backend/python/vllm \
     ; fi
+RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
+	PATH=$PATH:/opt/conda/bin make -C backend/python/mamba \
+    ; fi
 RUN if [ "${IMAGE_TYPE}" = "extras" ]; then \
 	PATH=$PATH:/opt/conda/bin make -C backend/python/sentencetransformers \
     ; fi

Makefile

@@ -419,6 +419,7 @@ protogen-python:
 	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/vall-e-x/ --grpc_python_out=backend/python/vall-e-x/ backend/backend.proto
 	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/vllm/ --grpc_python_out=backend/python/vllm/ backend/backend.proto
 	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/petals/ --grpc_python_out=backend/python/petals/ backend/backend.proto
+	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/mamba/ --grpc_python_out=backend/python/mamba/ backend/backend.proto
 	python3 -m grpc_tools.protoc -Ibackend/ --python_out=backend/python/exllama2/ --grpc_python_out=backend/python/exllama2/ backend/backend.proto
 
 ## GRPC
@@ -429,6 +430,7 @@ prepare-extra-conda-environments:
 	$(MAKE) -C backend/python/coqui
 	$(MAKE) -C backend/python/diffusers
 	$(MAKE) -C backend/python/vllm
+	$(MAKE) -C backend/python/mamba
 	$(MAKE) -C backend/python/sentencetransformers
 	$(MAKE) -C backend/python/transformers
 	$(MAKE) -C backend/python/transformers-musicgen

backend/backend.proto

@@ -112,7 +112,6 @@ message ModelOptions {
   int32 CLIPSkip = 33;
   string ControlNet = 48;
 
-  // RWKV
   string Tokenizer = 34;
 
   // LLM (llama.cpp)

backend/python/coqui/coqui_server.py

@@ -33,7 +33,13 @@ def Health(self, request, context):
     def LoadModel(self, request, context):
 
         # Get device
-        device = "cuda" if request.CUDA else "cpu"
+        # device = "cuda" if request.CUDA else "cpu"
+        if torch.cuda.is_available():
+            print("CUDA is available", file=sys.stderr)
+            device = "cuda"
+        else:
+            print("CUDA is not available", file=sys.stderr)
+            device = "cpu"
 
         if not torch.cuda.is_available() and request.CUDA:
             return backend_pb2.Result(success=False, message="CUDA is not available")

backend/python/mamba/backend_mamba.py

@@ -10,12 +10,16 @@
 import backend_pb2_grpc
 
 import grpc
-from vllm import LLM, SamplingParams
+
+import torch
+from transformers import AutoTokenizer, AutoModelForCausalLM
+from mamba_ssm.models.mixer_seq_simple import MambaLMHeadModel
 
 _ONE_DAY_IN_SECONDS = 60 * 60 * 24
 
 # If MAX_WORKERS are specified in the environment use it, otherwise default to 1
 MAX_WORKERS = int(os.environ.get('PYTHON_GRPC_MAX_WORKERS', '1'))
+MAMBA_CHAT= os.environ.get('MAMBA_CHAT', '1') == '1'
 
 # Implement the BackendServicer class with the service methods
 class BackendServicer(backend_pb2_grpc.BackendServicer):
@@ -80,10 +84,16 @@ def LoadModel(self, request, context):
             backend_pb2.Result: The load model result.
         """
         try:
-            if request.Quantization != "":
-                self.llm = LLM(model=request.Model, quantization=request.Quantization)
-            else:
-                self.llm = LLM(model=request.Model)
+            tokenizerModel = request.Tokenizer
+            if tokenizerModel == "":
+                tokenizerModel = request.Model
+
+            tokenizer = AutoTokenizer.from_pretrained(tokenizerModel)
+            if MAMBA_CHAT:
+                tokenizer.eos_token = "<|endoftext|>"
+                tokenizer.pad_token = tokenizer.eos_token
+            self.tokenizer = tokenizer
+            self.model = MambaLMHeadModel.from_pretrained(request.Model, device="cuda", dtype=torch.float16)
         except Exception as err:
             return backend_pb2.Result(success=False, message=f"Unexpected {err=}, {type(err)=}")
         return backend_pb2.Result(message="Model loaded successfully", success=True)
@@ -102,10 +112,20 @@ def Predict(self, request, context):
         if request.TopP == 0:
             request.TopP = 0.9
 
-        sampling_params = SamplingParams(temperature=request.Temperature, top_p=request.TopP)
-        outputs = self.llm.generate([request.Prompt], sampling_params)
+        max_tokens = request.Tokens
+
+        if request.Tokens == 0:
+            max_tokens = 2000
+
+        encoded_input = self.tokenizer(request.Prompt)
+
+        out = self.model.generate(input_ids=encoded_input["input_ids"], max_length=max_tokens, temperature=request.Temperratur,
+                                     top_p=request.TopP, eos_token_id=self.tokenizer.eos_token_id)
+
+        decoded = self.tokenizer.batch_decode(out)
+
+        generated_text = decoded[0]
 
-        generated_text = outputs[0].outputs[0].text
         # Remove prompt from response if present
         if request.Prompt in generated_text:
             generated_text = generated_text.replace(request.Prompt, "")