Update TensorRT-LLM

[kaiyux]

• Model Support
  • Support Grok-1, see examples/grok/README.md.
  • Support Qwen1.5-110B with FP8 PTQ.
  • Support Phi-3 small model with block sparse attention.
  • Support InternLM2 7B/20B, thanks to the contribution from @RunningLeon in Support internlm2 #1392.
• Features
  • Add a fused GEMM-SwiGLU plugin for FP8 on SM90.
  • Support HuggingFace model automatically download for the Python high level API.
  • Support running FP8 LLaMA with FP16 LoRA checkpoints.
  • Support very long context for LLaMA (see “Long context evaluation” section in examples/llama/README.md).
  • Support encoder-decoder models for C++ runtime with paged KV cache and inflight batching. enc-dec triton backend support #800
• Bug fixes
  • Fix qkv_bias shape issue for Qwen1.5-32B (convert qwen 110b gptq checkpoint的时候，qkv_bias 的shape不能被3整除 #1589), thanks to the contribution from @Tlntin in fix up qkv.bias error when use qwen1.5-32b-gptq-int4 #1637.
  • Fix the error of Ada traits for fpA_intB, thanks to the contribution from @JamesTheZ in Fix the error of Ada traits for fpA_intB. #1583.
  • Update examples/qwenvl/requirements.txt, thanks to the contribution from @ngoanpv in Update requirements.txt #1248.
  • Fix rsLoRA scaling in lora_manager, thanks to the contribution from @TheCodeWrangler in Fixed rslora scaling in lora_manager #1669.
  • Fix Qwen1.5 checkpoint convert failure convert_checkpoint qwen1.5 error #1675.
  • Fix Medusa safetensors and AWQ conversion, thanks to the contribution from @Tushar-ml in Loading Medusa Safetensors + AWQ Conversion correction #1535.
• Documentation
  • Add HuggingFace model zoo from the community, thanks to the contribution from @matichon-vultureprime in Add Huggingface model zoo from community #1674.
  • Fix benchmarks/cpp/README.md for gptManagerBenchmark seems to go into a dead loop with GPU usage 0% #1562 and Cannot process new request: [TensorRT-LLM][ERROR] Assertion failed: LoRA task 0 not found in cache. Please send LoRA weights with request (/home/jenkins/agent/workspace/LLM/main/L0_PostMerge/llm/cpp/tensorrt_llm/batch_manager/peftCacheManager.cpp:182) #1552.
  • Fix dead link, thanks to the help from @DefTruth, @buvnswrn and @sunjiabin17 in: [Docs] Fixed inference-request.md dead link triton-inference-server/tensorrtllm_backend#478, Fixed README.md for broken links triton-inference-server/tensorrtllm_backend#482 and FIX link reference in README.md triton-inference-server/tensorrtllm_backend#449.