Update Kimi K2.5 recipes: container, model path, concurrency format, and env cleanup

- Update container to tensorrtllm-runtime-1.1.0-dev.2.sqsh
- Point model path to shared /mnt/lustre01/models/kimi-k2.5-nvfp4
- Update Eagle model mount path for MTP configs
- Remove HF_HOME (defaults to ~/.cache/huggingface)
- Fix concurrency separator from space to 'x' for sa-bench compatibility
- Enable multiple frontends for ctx1dep4_gen1dep32_batch64

Author: yeswanthk-26

recipes/kimi2.5/trtllm_dynamo/disagg/gb200Nvfp4/ISL1K_OSL1K/MTP/ctx1dep4_gen1dep16_batch32_eplb0_mtp3.yaml

@@ -6,8 +6,8 @@ name: "kimi_k25_nvfp4_ISL1K_OSL1K_ctx1dep4_gen1dep16_batch32_eplb0_mtp3"
 # concurrency: 666
 
 model:
-  path: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-NVFP4"
-  container: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/squash/trtllm-main_aarch-46939060.sqsh"
+  path: "/mnt/lustre01/models/kimi-k2.5-nvfp4"
+  container: "/mnt/lustre01/users/slurm-shared/yeswanthk/squashs/tensorrtllm-runtime-1.1.0-dev.2.sqsh"
   precision: "fp4"
 
 resources:
@@ -33,7 +33,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   decode_environment:
     ENROOT_ALLOW_DEV: "yes"
@@ -43,7 +42,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   trtllm_config:
     prefill:
@@ -117,7 +115,7 @@ backend:
         speculative_model_dir: "/eagle-model"
 
 extra_mount:
-  - "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
+  - "/mnt/lustre01/users/slurm-shared/yeswanthk/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
 
 benchmark:
   type: "sa-bench"

recipes/kimi2.5/trtllm_dynamo/disagg/gb200Nvfp4/ISL1K_OSL1K/MTP/ctx1dep4_gen1dep32_batch16_eplb0_mtp3.yaml

@@ -6,8 +6,8 @@ name: "kimi_k25_nvfp4_ISL1K_OSL1K_ctx1dep4_gen1dep32_batch16_eplb0_mtp3"
 # concurrency: 666
 
 model:
-  path: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-NVFP4"
-  container: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/squash/trtllm-main_aarch-46939060.sqsh"
+  path: "/mnt/lustre01/models/kimi-k2.5-nvfp4"
+  container: "/mnt/lustre01/users/slurm-shared/yeswanthk/squashs/tensorrtllm-runtime-1.1.0-dev.2.sqsh"
   precision: "fp4"
 
 resources:
@@ -33,7 +33,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   decode_environment:
     ENROOT_ALLOW_DEV: "yes"
@@ -43,7 +42,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   trtllm_config:
     prefill:
@@ -115,7 +113,7 @@ backend:
         speculative_model_dir: "/eagle-model"
 
 extra_mount:
-  - "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
+  - "/mnt/lustre01/users/slurm-shared/yeswanthk/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
 
 benchmark:
   type: "sa-bench"

recipes/kimi2.5/trtllm_dynamo/disagg/gb200Nvfp4/ISL1K_OSL1K/MTP/ctx1dep4_gen1dep8_batch512_eplb0_mtp1.yaml

@@ -6,8 +6,8 @@ name: "kimi_k25_nvfp4_ISL1K_OSL1K_ctx1dep4_gen1dep8_batch512_eplb0_mtp1"
 # concurrency: 4301
 
 model:
-  path: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-NVFP4"
-  container: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/squash/trtllm-main_aarch-46939060.sqsh"
+  path: "/mnt/lustre01/models/kimi-k2.5-nvfp4"
+  container: "/mnt/lustre01/users/slurm-shared/yeswanthk/squashs/tensorrtllm-runtime-1.1.0-dev.2.sqsh"
   precision: "fp4"
 
 resources:
@@ -33,7 +33,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   decode_environment:
     ENROOT_ALLOW_DEV: "yes"
@@ -43,7 +42,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   trtllm_config:
     prefill:
@@ -177,7 +175,7 @@ backend:
         speculative_model_dir: "/eagle-model"
 
 extra_mount:
-  - "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
+  - "/mnt/lustre01/users/slurm-shared/yeswanthk/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
 
 benchmark:
   type: "sa-bench"

recipes/kimi2.5/trtllm_dynamo/disagg/gb200Nvfp4/ISL1K_OSL1K/MTP/ctx1dep4_gen4tep8_batch64_allconc_eplb0_mtp3.yaml

@@ -6,8 +6,8 @@ name: "kimi_k25_nvfp4_ISL1K_OSL1K_ctx1dep4_gen4tep8_batch64_allconc_eplb0_mtp3"
 # Covers all gen4tep8 concurrencies: 8, 48, 92, 192, 336
 
 model:
-  path: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-NVFP4"
-  container: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/squash/trtllm-main_aarch-46939060.sqsh"
+  path: "/mnt/lustre01/models/kimi-k2.5-nvfp4"
+  container: "/mnt/lustre01/users/slurm-shared/yeswanthk/squashs/tensorrtllm-runtime-1.1.0-dev.2.sqsh"
   precision: "fp4"
 
 resources:
@@ -33,7 +33,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   decode_environment:
     ENROOT_ALLOW_DEV: "yes"
@@ -43,7 +42,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   trtllm_config:
     prefill:
@@ -122,13 +120,13 @@ backend:
         speculative_model_dir: "/eagle-model"
 
 extra_mount:
-  - "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
+  - "/mnt/lustre01/users/slurm-shared/yeswanthk/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
 
 benchmark:
   type: "sa-bench"
   isl: 1024
   osl: 1024
-  concurrencies: "8 48 92 192 336"
+  concurrencies: "8x48x92x192x336"
   req_rate: "inf"
 
 frontend:

recipes/kimi2.5/trtllm_dynamo/disagg/gb200Nvfp4/ISL1K_OSL1K/MTP/ctx1dep4_gen5tep4_batch2_allconc_eplb0_mtp3.yaml

@@ -6,8 +6,8 @@ name: "kimi_k25_nvfp4_ISL1K_OSL1K_ctx1dep4_gen5tep4_batch2_allconc_eplb0_mtp3"
 # Covers all gen5tep4 concurrencies: 10, 15
 
 model:
-  path: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-NVFP4"
-  container: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/squash/trtllm-main_aarch-46939060.sqsh"
+  path: "/mnt/lustre01/models/kimi-k2.5-nvfp4"
+  container: "/mnt/lustre01/users/slurm-shared/yeswanthk/squashs/tensorrtllm-runtime-1.1.0-dev.2.sqsh"
   precision: "fp4"
 
 resources:
@@ -33,7 +33,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   decode_environment:
     ENROOT_ALLOW_DEV: "yes"
@@ -43,7 +42,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   trtllm_config:
     prefill:
@@ -113,13 +111,13 @@ backend:
         speculative_model_dir: "/eagle-model"
 
 extra_mount:
-  - "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
+  - "/mnt/lustre01/users/slurm-shared/yeswanthk/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
 
 benchmark:
   type: "sa-bench"
   isl: 1024
   osl: 1024
-  concurrencies: "10 15"
+  concurrencies: "10x15"
   req_rate: "inf"
 
 frontend:

recipes/kimi2.5/trtllm_dynamo/disagg/gb200Nvfp4/ISL1K_OSL1K/MTP/ctx2dep4_gen1dep16_batch128_eplb0_mtp1.yaml

@@ -6,8 +6,8 @@ name: "kimi_k25_nvfp4_ISL1K_OSL1K_ctx2dep4_gen1dep16_batch128_eplb0_mtp1"
 # concurrency: 2253
 
 model:
-  path: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-NVFP4"
-  container: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/squash/trtllm-main_aarch-46939060.sqsh"
+  path: "/mnt/lustre01/models/kimi-k2.5-nvfp4"
+  container: "/mnt/lustre01/users/slurm-shared/yeswanthk/squashs/tensorrtllm-runtime-1.1.0-dev.2.sqsh"
   precision: "fp4"
 
 resources:
@@ -33,7 +33,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   decode_environment:
     ENROOT_ALLOW_DEV: "yes"
@@ -43,7 +42,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   trtllm_config:
     prefill:
@@ -129,7 +127,7 @@ backend:
         speculative_model_dir: "/eagle-model"
 
 extra_mount:
-  - "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
+  - "/mnt/lustre01/users/slurm-shared/yeswanthk/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
 
 benchmark:
   type: "sa-bench"

recipes/kimi2.5/trtllm_dynamo/disagg/gb200Nvfp4/ISL1K_OSL1K/MTP/ctx2dep4_gen1dep32_batch64_eplb0_mtp1.yaml

@@ -6,8 +6,8 @@ name: "kimi_k25_nvfp4_ISL1K_OSL1K_ctx2dep4_gen1dep32_batch64_eplb0_mtp1"
 # concurrency: 2253
 
 model:
-  path: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-NVFP4"
-  container: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/squash/trtllm-main_aarch-46939060.sqsh"
+  path: "/mnt/lustre01/models/kimi-k2.5-nvfp4"
+  container: "/mnt/lustre01/users/slurm-shared/yeswanthk/squashs/tensorrtllm-runtime-1.1.0-dev.2.sqsh"
   precision: "fp4"
 
 resources:
@@ -33,7 +33,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   decode_environment:
     ENROOT_ALLOW_DEV: "yes"
@@ -43,7 +42,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   trtllm_config:
     prefill:
@@ -121,7 +119,7 @@ backend:
         speculative_model_dir: "/eagle-model"
 
 extra_mount:
-  - "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
+  - "/mnt/lustre01/users/slurm-shared/yeswanthk/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
 
 benchmark:
   type: "sa-bench"

recipes/kimi2.5/trtllm_dynamo/disagg/gb200Nvfp4/ISL1K_OSL1K/MTP/ctx2dep4_gen3dep8_batch256_eplb0_mtp1.yaml

@@ -6,8 +6,8 @@ name: "kimi_k25_nvfp4_ISL1K_OSL1K_ctx2dep4_gen3dep8_batch256_eplb0_mtp1"
 # concurrency: 6759
 
 model:
-  path: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-NVFP4"
-  container: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/squash/trtllm-main_aarch-46939060.sqsh"
+  path: "/mnt/lustre01/models/kimi-k2.5-nvfp4"
+  container: "/mnt/lustre01/users/slurm-shared/yeswanthk/squashs/tensorrtllm-runtime-1.1.0-dev.2.sqsh"
   precision: "fp4"
 
 resources:
@@ -33,7 +33,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   decode_environment:
     ENROOT_ALLOW_DEV: "yes"
@@ -43,7 +42,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   trtllm_config:
     prefill:
@@ -145,7 +143,7 @@ backend:
         speculative_model_dir: "/eagle-model"
 
 extra_mount:
-  - "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
+  - "/mnt/lustre01/users/slurm-shared/yeswanthk/models/Kimi-K2.5-Thinking-Eagle3:/eagle-model"
 
 benchmark:
   type: "sa-bench"

recipes/kimi2.5/trtllm_dynamo/disagg/gb200Nvfp4/ISL1K_OSL1K/STP/ctx1dep4_gen1dep16_batch32_eplb0_mtp0.yaml

@@ -6,8 +6,8 @@ name: "kimi_k25_nvfp4_ISL1K_OSL1K_ctx1dep4_gen1dep16_batch32_eplb0_mtp0"
 # concurrency: 666
 
 model:
-  path: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-NVFP4"
-  container: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/squash/trtllm-main_aarch-46939060.sqsh"
+  path: "/mnt/lustre01/models/kimi-k2.5-nvfp4"
+  container: "/mnt/lustre01/users/slurm-shared/yeswanthk/squashs/tensorrtllm-runtime-1.1.0-dev.2.sqsh"
   precision: "fp4"
 
 resources:
@@ -33,7 +33,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   decode_environment:
     ENROOT_ALLOW_DEV: "yes"
@@ -43,7 +42,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   trtllm_config:
     prefill:

recipes/kimi2.5/trtllm_dynamo/disagg/gb200Nvfp4/ISL1K_OSL1K/STP/ctx1dep4_gen1dep32_batch64_eplb0_mtp0.yaml

@@ -6,8 +6,8 @@ name: "kimi_k25_nvfp4_ISL1K_OSL1K_ctx1dep4_gen1dep32_batch64_eplb0_mtp0"
 # concurrency: 2253
 
 model:
-  path: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/models/Kimi-K2.5-NVFP4"
-  container: "/lustre/fsw/infra_rd_gsw/users/yeswanthk/srt-slurm/squash/trtllm-main_aarch-46939060.sqsh"
+  path: "/mnt/lustre01/models/kimi-k2.5-nvfp4"
+  container: "/mnt/lustre01/users/slurm-shared/yeswanthk/squashs/tensorrtllm-runtime-1.1.0-dev.2.sqsh"
   precision: "fp4"
 
 resources:
@@ -33,7 +33,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   decode_environment:
     ENROOT_ALLOW_DEV: "yes"
@@ -43,7 +42,6 @@ backend:
     TRTLLM_ENABLE_PDL: "1"
     TRTLLM_SERVER_DISABLE_GC: "1"
     TRTLLM_WORKER_DISABLE_GC: "1"
-    HF_HOME: "/lustre/fsw/coreai_comparch_infbench/common/hf_cache"
 
   trtllm_config:
     prefill:
@@ -121,7 +119,7 @@ benchmark:
 
 frontend:
   type: "dynamo"
-  enable_multiple_frontends: false
+  enable_multiple_frontends: true
 
 health_check:
   max_attempts: 360