WARNING: All log messages before absl::InitializeLog() is called are written to STDERR I0000 00:00:1777098154.146342 72481 port.cc:153] oneDNN custom operations are on. You may see slightly different numerical results due to floating-point round-off errors from different computation orders. To turn them off, set the environment variable `TF_ENABLE_ONEDNN_OPTS=0`. Skipping import of cpp extensions due to incompatible torch version. Please upgrade to torch >= 2.11.0 (found 2.10.0+cu128). ============== Export Configuration ============== auto_model_override : None batch_size : 1 bundle_litert_lm : True cache_implementation : 'LiteRTLMCache' cache_length : 1024 cache_length_dim : None enable_dynamic_shape : False experimental_lightweight_conversion : False experimental_use_mixed_precision : False export_vision_encoder : False externalize_embedder : True externalize_rope : False extra_kwargs : {} k_ts_idx : 2 keep_temporary_files : False litert_lm_llm_metadata_override : None litert_lm_model_type_override : None model : 'punjabi_gemma/ankahi' output_dir : 'artifacts/litertlm' prefill_length_dim : None prefill_lengths : [512] quantization_recipe : 'int8' single_token_embedder : False split_cache : False task : trust_remote_code : True use_jinja_template : True v_ts_idx : 3 vision_encoder_quantization_recipe : 'weight_only_wi8_afp32' work_dir : 'artifacts/litertlm/tmpn60v351f' ================================================== (00:00) [START] LiteRT GenAI Export (00:00) [START] LiteRT GenAI Export > Load source model Gemma4 patch applied. Loading weights: 0%| | 0/2130 [00:00 Load source model (+00:20) (00:20) [START] LiteRT GenAI Export > Export text prefill-decode model Using Gemma4 exportables. (00:21) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert (00:21) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Torch Export: prefill_512 (00:31) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Torch Export: prefill_512 > ExportedProgram Run Decompositions (00:48) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Torch Export: prefill_512 > ExportedProgram Run Decompositions (+00:17) (00:48) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Torch Export: prefill_512 (+00:27) (00:48) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Torch Export: decode (00:58) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Torch Export: decode > ExportedProgram Run Decompositions (01:15) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Torch Export: decode > ExportedProgram Run Decompositions (+00:17) (01:15) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Torch Export: decode (+00:27) (01:15) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Run FX Passes (01:16) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Run FX Passes > ExportedProgram Run Decompositions (01:16) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Run FX Passes > ExportedProgram Run Decompositions (+00:00) (01:17) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Run FX Passes > ExportedProgram Run Decompositions (01:17) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Run FX Passes > ExportedProgram Run Decompositions (+00:00) (01:18) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Run FX Passes (+00:02) (01:18) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: prefill_512 (01:18) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: prefill_512 > ExportedProgram Run Decompositions (01:32) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: prefill_512 > ExportedProgram Run Decompositions (+00:14) (01:32) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: prefill_512 > ExportedProgram Run Decompositions (01:32) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: prefill_512 > ExportedProgram Run Decompositions (+00:00) (01:32) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: prefill_512 > Create MLIR Module WARNING:2026-04-25 06:24:57,531:jax._src.xla_bridge:794: An NVIDIA GPU may be present on this machine, but a CUDA-enabled jaxlib is not installed. Falling back to cpu. WARNING:jax._src.xla_bridge:An NVIDIA GPU may be present on this machine, but a CUDA-enabled jaxlib is not installed. Falling back to cpu. (02:30) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: prefill_512 > Create MLIR Module (+00:57) (02:30) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: prefill_512 (+01:12) (02:30) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: decode (02:30) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: decode > ExportedProgram Run Decompositions (02:49) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: decode > ExportedProgram Run Decompositions (+00:19) (02:49) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: decode > ExportedProgram Run Decompositions (02:49) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: decode > ExportedProgram Run Decompositions (+00:00) (02:49) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: decode > Create MLIR Module (03:04) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: decode > Create MLIR Module (+00:14) (03:04) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Lower to MLIR: decode (+00:33) (03:04) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Merge MLIR Modules (03:04) [ DONE] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Merge MLIR Modules (+00:00) (03:04) [START] LiteRT GenAI Export > Export text prefill-decode model > LiteRT-Torch Convert > Run LiteRT Converter Passes