Add TRT-LLM converted checkpoint (ready for build)

.gitattributes

@@ -34,3 +34,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
 tokenizer.json filter=lfs diff=lfs merge=lfs -text
+trtllm_checkpoint/tokenizer.json filter=lfs diff=lfs merge=lfs -text

trtllm_checkpoint/chat_template.jinja

@@ -0,0 +1,3 @@
+{% if messages[0]['role'] == 'system' %}{% set system_message = messages[0]['content'] | trim + '
+
+' %}{% set messages = messages[1:] %}{% else %}{% set system_message = '' %}{% endif %}{{- bos_token + system_message}}{% for message in messages %}{% if message['role'] == 'user' %}{{ '[INST]' + message['content'] | trim + '[/INST]' }}{% elif message['role'] == 'assistant' %}{{ message['content'] | trim + eos_token }}{% endif %}{% endfor %}

trtllm_checkpoint/config.json

@@ -0,0 +1,90 @@
+{
+    "producer": {
+        "name": "modelopt",
+        "version": "0.37.0"
+    },
+    "architecture": "MistralForCausalLM",
+    "dtype": "bfloat16",
+    "logits_dtype": "float16",
+    "num_hidden_layers": 40,
+    "num_attention_heads": 32,
+    "num_key_value_heads": 8,
+    "hidden_size": 5120,
+    "norm_epsilon": 1e-05,
+    "vocab_size": 131072,
+    "max_position_embeddings": 1024000,
+    "hidden_act": "silu",
+    "use_parallel_embedding": true,
+    "embedding_sharding_dim": 0,
+    "head_size": 128,
+    "intermediate_size": 14336,
+    "position_embedding_type": "rope_gpt_neox",
+    "share_embedding_table": false,
+    "residual_mlp": false,
+    "bias": false,
+    "rotary_pct": 1.0,
+    "rank": 0,
+    "decoder": "mistral",
+    "rmsnorm": true,
+    "lm_head_bias": false,
+    "mlp_bias": false,
+    "attn_bias": false,
+    "rotary_base": 1000000.0,
+    "rotary_scaling": null,
+    "disable_weight_only_quant_plugin": false,
+    "moe": {
+        "num_experts": 0,
+        "shared_expert_intermediate_size": 0,
+        "top_k": 0,
+        "normalization_mode": null,
+        "sparse_mixer_epsilon": 0.01,
+        "tp_mode": 0,
+        "device_limited_n_group": 0,
+        "device_limited_topk_group": 0,
+        "device_limited_routed_scaling_factor": 1.0
+    },
+    "remove_duplicated_kv_heads": false,
+    "fc_after_embed": false,
+    "use_input_layernorm_in_first_layer": true,
+    "use_last_layernorm": true,
+    "layer_idx_offset": 0,
+    "embedding_multiplier": 1.0,
+    "attention_multiplier": 1.0,
+    "residual_multiplier": 1.0,
+    "output_multiplier_scale": 1.0,
+    "has_partial_lora_mask": false,
+    "runtime_defaults": null,
+    "mapping": {
+        "world_size": 1,
+        "gpus_per_node": 8,
+        "cp_size": 1,
+        "tp_size": 1,
+        "pp_size": 1,
+        "moe_tp_size": 1,
+        "moe_cluster_size": 1,
+        "moe_ep_size": 1,
+        "attn_tp_size": 1,
+        "attn_cp_size": 1,
+        "cp_config": {},
+        "auto_parallel": false,
+        "enable_attention_dp": false,
+        "enable_lm_head_tp_in_adp": false
+    },
+    "quantization": {
+        "quant_algo": "NVFP4",
+        "kv_cache_quant_algo": null,
+        "group_size": 16,
+        "smoothquant_val": 0.5,
+        "clamp_val": null,
+        "use_meta_recipe": false,
+        "has_zero_point": false,
+        "pre_quant_scale": false,
+        "exclude_modules": [
+            "lm_head"
+        ],
+        "mamba_ssm_cache_dtype": null
+    },
+    "qk_layernorm": false,
+    "rotary_embedding_dim": 128,
+    "tie_word_embeddings": false
+}

trtllm_checkpoint/rank0.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:464bcb3334684b3061d00c016b267ce129f72c11a103ad7569b9969bd8e9c0cc
+size 8819460504

trtllm_checkpoint/special_tokens_map.json

@@ -0,0 +1,30 @@
+{
+  "bos_token": {
+    "content": "<s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "</s>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "<pad>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "unk_token": {
+    "content": "<unk>",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

trtllm_checkpoint/tokenizer.json

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:8771d3c14b4206fd6b54c60d037eeafce2f37382046c2369c36cd2edd2f099d7
+size 17078391