Upload folder using huggingface_hub

config.json

@@ -0,0 +1,19 @@
+{
+  "architectures": [
+    "LlamaForCausalLM"
+  ],
+  "max_position_embeddings": 16384,
+  "model_type": "llama",
+  "num_hidden_layers": 30,
+  "hidden_size": 4096,
+  "intermediate_size": 11008,
+  "head_dim": 128,
+  "num_attention_heads": 32,
+  "num_key_value_heads": 32,  
+  "rms_norm_eps": 1e-06,
+  "rope_theta": 10000.0,
+  "tie_word_embeddings": false,
+  "torch_dtype": "bfloat16",
+  "vocab_size": 102400,
+  "transformers_version": "4.33.1"
+}

model.safetensors

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e7849160fbd3cbf16c2553345ee599bc7054cfb5bd899c4fa7c201e66adaa19a
+size 13820763056

preprocessor_config.json

@@ -0,0 +1,23 @@
+{
+  "background_color": [
+    127,
+    127,
+    127
+  ],
+  "do_normalize": true,
+  "image_mean": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "image_processor_type": "VLMImageProcessor",
+  "image_size": 384,
+  "image_std": [
+    0.5,
+    0.5,
+    0.5
+  ],
+  "min_size": 14,
+  "processor_class": "VLChatProcessor",
+  "rescale_factor": 0.00392156862745098
+}

processor_config.json

@@ -0,0 +1,9 @@
+{
+  "add_special_token": false,
+  "ignore_id": -100,
+  "image_tag": "<image_placeholder>",
+  "mask_prompt": true,
+  "num_image_tokens": 576,
+  "processor_class": "VLChatProcessor",
+  "sft_format": "deepseek"
+}

pytorch_model.bin.index.json

@@ -0,0 +1,280 @@
+{
+  "metadata": {
+    "total_size": 14840868118
+  },
+  "weight_map": {
+    "lm_head.weight": "pytorch_model.bin",
+    "model.embed_tokens.weight": "pytorch_model.bin",
+    "model.layers.0.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.0.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.0.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.0.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.0.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.0.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.0.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.0.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.0.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.1.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.1.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.1.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.1.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.1.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.1.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.1.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.1.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.1.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.10.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.10.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.10.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.10.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.10.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.10.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.10.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.10.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.10.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.11.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.11.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.11.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.11.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.11.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.11.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.11.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.11.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.11.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.12.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.12.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.12.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.12.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.12.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.12.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.12.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.12.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.12.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.13.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.13.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.13.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.13.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.13.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.13.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.13.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.13.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.13.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.14.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.14.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.14.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.14.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.14.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.14.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.14.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.14.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.14.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.15.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.15.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.15.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.15.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.15.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.15.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.15.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.15.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.15.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.16.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.16.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.16.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.16.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.16.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.16.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.16.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.16.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.16.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.17.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.17.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.17.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.17.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.17.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.17.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.17.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.17.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.17.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.18.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.18.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.18.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.18.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.18.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.18.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.18.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.18.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.18.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.19.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.19.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.19.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.19.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.19.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.19.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.19.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.19.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.19.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.2.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.2.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.2.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.2.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.2.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.2.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.2.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.2.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.2.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.20.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.20.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.20.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.20.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.20.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.20.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.20.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.20.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.20.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.21.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.21.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.21.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.21.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.21.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.21.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.21.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.21.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.21.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.22.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.22.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.22.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.22.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.22.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.22.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.22.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.22.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.22.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.23.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.23.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.23.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.23.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.23.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.23.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.23.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.23.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.23.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.24.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.24.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.24.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.24.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.24.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.24.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.24.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.24.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.24.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.25.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.25.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.25.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.25.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.25.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.25.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.25.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.25.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.25.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.26.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.26.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.26.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.26.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.26.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.26.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.26.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.26.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.26.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.27.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.27.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.27.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.27.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.27.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.27.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.27.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.27.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.27.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.28.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.28.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.28.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.28.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.28.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.28.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.28.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.28.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.28.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.29.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.29.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.29.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.29.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.29.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.29.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.29.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.29.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.29.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.3.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.3.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.3.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.3.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.3.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.3.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.3.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.3.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.3.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.4.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.4.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.4.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.4.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.4.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.4.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.4.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.4.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.4.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.5.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.5.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.5.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.5.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.5.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.5.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.5.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.5.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.5.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.6.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.6.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.6.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.6.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.6.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.6.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.6.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.6.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.6.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.7.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.7.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.7.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.7.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.7.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.7.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.7.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.7.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.7.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.8.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.8.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.8.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.8.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.8.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.8.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.8.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.8.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.8.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.layers.9.input_layernorm.weight": "pytorch_model.bin",
+    "model.layers.9.mlp.down_proj.weight": "pytorch_model.bin",
+    "model.layers.9.mlp.gate_proj.weight": "pytorch_model.bin",
+    "model.layers.9.mlp.up_proj.weight": "pytorch_model.bin",
+    "model.layers.9.post_attention_layernorm.weight": "pytorch_model.bin",
+    "model.layers.9.self_attn.k_proj.weight": "pytorch_model.bin",
+    "model.layers.9.self_attn.o_proj.weight": "pytorch_model.bin",
+    "model.layers.9.self_attn.q_proj.weight": "pytorch_model.bin",
+    "model.layers.9.self_attn.v_proj.weight": "pytorch_model.bin",
+    "model.norm.weight": "pytorch_model.bin"
+  }
+}

special_tokens_map.json

@@ -0,0 +1,16 @@
+{
+  "additional_special_tokens": [
+    "<image_placeholder>",
+    "<patch_placeholder>",
+    "<|ref|>",
+    "<|/ref|>",
+    "<|det|>",
+    "<|/det|>",
+    "<|grounding|>",
+    "<|User|>",
+    "<|Assistant|>"
+  ],
+  "bos_token": "<｜begin▁of▁sentence｜>",
+  "eos_token": "<｜end▁of▁sentence｜>",
+  "pad_token": "<｜▁pad▁｜>"
+}

test_conv.py

@@ -0,0 +1,365 @@
+# Copyright (c) 2023-2024 DeepSeek.
+#
+# Permission is hereby granted, free of charge, to any person obtaining a copy of
+# this software and associated documentation files (the "Software"), to deal in
+# the Software without restriction, including without limitation the rights to
+# use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of
+# the Software, and to permit persons to whom the Software is furnished to do so,
+# subject to the following conditions:
+#
+# The above copyright notice and this permission notice shall be included in all
+# copies or substantial portions of the Software.
+#
+# THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+# IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS
+# FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR
+# COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER
+# IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN
+# CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
+
+"""
+From https://github.com/lm-sys/FastChat/blob/main/fastchat/conversation.py
+"""
+
+import dataclasses
+from enum import IntEnum, auto
+from typing import Dict, List
+
+
+class SeparatorStyle(IntEnum):
+    """Separator styles."""
+
+    ADD_COLON_SINGLE = auto()
+    ADD_COLON_TWO = auto()
+    ADD_COLON_SPACE_SINGLE = auto()
+    NO_COLON_SINGLE = auto()
+    NO_COLON_TWO = auto()
+    ADD_NEW_LINE_SINGLE = auto()
+    LLAMA2 = auto()
+    CHATGLM = auto()
+    CHATML = auto()
+    CHATINTERN = auto()
+    DOLLY = auto()
+    RWKV = auto()
+    PHOENIX = auto()
+    ROBIN = auto()
+    DeepSeek = auto()
+    PLAIN = auto()
+    ALIGNMENT = auto()
+
+
+@dataclasses.dataclass
+class Conversation:
+    """A class that manages prompt templates and keeps all conversation history."""
+
+    # The name of this template
+    name: str
+    # The template of the system prompt
+    system_template: str = "{system_message}"
+    # The system message
+    system_message: str = ""
+    # The names of two roles
+    roles: List[str] = (("USER", "ASSISTANT"),)
+    # All messages. Each item is (role, message).
+    messages: List[List[str]] = ()
+    # The number of few shot examples
+    offset: int = 0
+    # The separator style and configurations
+    sep_style: SeparatorStyle = SeparatorStyle.ADD_COLON_SINGLE
+    sep: str = "\n"
+    sep2: str = None
+    # Stop criteria (the default one is EOS token)
+    stop_str: str = None
+    # Stops generation if meeting any token in this list
+    stop_token_ids: List[int] = None
+
+    def get_prompt(self) -> str:
+        """Get the prompt for generation."""
+        system_prompt = self.system_template.format(system_message=self.system_message)
+
+        if self.sep_style == SeparatorStyle.DeepSeek:
+            seps = [self.sep, self.sep2]
+            if system_prompt == "" or system_prompt is None:
+                ret = ""
+            else:
+                ret = system_prompt + seps[0]
+            for i, (role, message) in enumerate(self.messages):
+                if message:
+                    ret += role + ": " + message + seps[i % 2]
+                else:
+                    ret += role + ":"
+            return ret
+        elif self.sep_style == SeparatorStyle.LLAMA2:
+            seps = [self.sep, self.sep2]
+            if self.system_message:
+                ret = system_prompt
+            else:
+                ret = "[INST] "
+            for i, (role, message) in enumerate(self.messages):
+                tag = self.roles[i % 2]
+                if message:
+                    if type(message) is tuple:  # multimodal message
+                        message, _ = message
+                    if i == 0:
+                        ret += message + " "
+                    else:
+                        ret += tag + " " + message + seps[i % 2]
+                else:
+                    ret += tag
+            return ret
+        elif self.sep_style == SeparatorStyle.PLAIN:
+            seps = [self.sep, self.sep2]
+            ret = ""
+            for i, (role, message) in enumerate(self.messages):
+                if message:
+                    if type(message) is tuple:
+                        message, _, _ = message
+                    if i % 2 == 0:
+                        ret += message + seps[i % 2]
+                    else:
+                        ret += message + seps[i % 2]
+                else:
+                    ret += ""
+            return ret
+        elif self.sep_style == SeparatorStyle.ALIGNMENT:
+            seps = [self.sep, self.sep2]
+            ret = ""
+            for i, (role, message) in enumerate(self.messages):
+                if message:
+                    if type(message) is tuple:
+                        message, _, _ = message
+                    if i % 2 == 0:
+                        ret += "<image>\n" + seps[i % 2]
+                    else:
+                        ret += message + seps[i % 2]
+                else:
+                    ret += ""
+            return ret
+        else:
+            raise ValueError(f"Invalid style: {self.sep_style}")
+
+    def get_prompt_for_current_round(self, content=None):
+        """Get current round formatted question prompt during sft training"""
+        if self.sep_style == SeparatorStyle.PLAIN:
+            formatted_question = "<image>\n"
+        elif self.sep_style == SeparatorStyle.DeepSeek:
+            formatted_question = (
+                f"{self.roles[0]}: " + content.strip() + self.sep + f"{self.roles[1]}:"
+            )
+        else:
+            raise ValueError(f"Unsupported sep_style: {self.sep_style}")
+        return formatted_question
+
+    def set_system_message(self, system_message: str):
+        """Set the system message."""
+        self.system_message = system_message
+
+    def append_message(self, role: str, message: str):
+        """Append a new message."""
+        self.messages.append([role, message])
+
+    def reset_message(self):
+        """Reset a new message."""
+        self.messages = []
+
+    def update_last_message(self, message: str):
+        """Update the last output.
+
+        The last message is typically set to be None when constructing the prompt,
+        so we need to update it in-place after getting the response from a model.
+        """
+        self.messages[-1][1] = message
+
+    def to_gradio_chatbot(self):
+        """Convert the conversation to gradio chatbot format."""
+        ret = []
+        for i, (role, msg) in enumerate(self.messages[self.offset :]):
+            if i % 2 == 0:
+                ret.append([msg, None])
+            else:
+                ret[-1][-1] = msg
+        return ret
+
+    def to_openai_api_messages(self):
+        """Convert the conversation to OpenAI chat completion format."""
+        system_prompt = self.system_template.format(system_message=self.system_message)
+        ret = [{"role": "system", "content": system_prompt}]
+
+        for i, (_, msg) in enumerate(self.messages[self.offset :]):
+            if i % 2 == 0:
+                ret.append({"role": "user", "content": msg})
+            else:
+                if msg is not None:
+                    ret.append({"role": "assistant", "content": msg})
+        return ret
+
+    def copy(self):
+        return Conversation(
+            name=self.name,
+            system_template=self.system_template,
+            system_message=self.system_message,
+            roles=self.roles,
+            messages=[[x, y] for x, y in self.messages],
+            offset=self.offset,
+            sep_style=self.sep_style,
+            sep=self.sep,
+            sep2=self.sep2,
+            stop_str=self.stop_str,
+            stop_token_ids=self.stop_token_ids,
+        )
+
+    def dict(self):
+        return {
+            "template_name": self.name,
+            "system_message": self.system_message,
+            "roles": self.roles,
+            "messages": self.messages,
+            "offset": self.offset,
+        }
+
+
+# A global registry for all conversation templates
+conv_templates: Dict[str, Conversation] = {}
+
+
+def register_conv_template(template: Conversation, override: bool = False):
+    """Register a new conversation template."""
+    if not override:
+        assert (
+            template.name not in conv_templates
+        ), f"{template.name} has been registered."
+
+    conv_templates[template.name] = template
+
+
+def get_conv_template(name: str) -> Conversation:
+    """Get a conversation template."""
+    return conv_templates[name].copy()
+
+
+# llava_llama2 template
+register_conv_template(
+    Conversation(
+        name="llava_llama2",
+        system_message="You are a helpful language and vision assistant. "
+        "You are able to understand the visual content that the user provides, "
+        "and assist the user with a variety of tasks using natural language.",
+        system_template="[INST] <<SYS>>\n{system_message}\n<</SYS>>\n\n",
+        roles=("[INST]", "[/INST]"),
+        messages=(),
+        offset=0,
+        sep_style=SeparatorStyle.LLAMA2,
+        sep=" ",
+        sep2=" </s><s>",
+        stop_token_ids=[2],
+    )
+)
+
+# llama2 template
+# reference: https://github.com/facebookresearch/llama/blob/cfc3fc8c1968d390eb830e65c63865e980873a06/llama/generation.py#L212
+register_conv_template(
+    Conversation(
+        name="llama-2",
+        system_template="[INST] <<SYS>>\n{system_message}\n<</SYS>>\n\n",
+        roles=("[INST]", "[/INST]"),
+        messages=(),
+        offset=0,
+        sep_style=SeparatorStyle.LLAMA2,
+        sep=" ",
+        sep2=" </s><s>",
+        stop_token_ids=[2],
+    )
+)
+
+
+# deepseek template
+register_conv_template(
+    Conversation(
+        name="deepseek_old",
+        system_template="{system_message}",
+        # system_message="You are a helpful assistant. Please answer truthfully and write out your "
+        # "thinking step by step to be sure you get the right answer.",
+        system_message="",
+        roles=("User", "Assistant"),
+        messages=(),
+        offset=0,
+        sep_style=SeparatorStyle.DeepSeek,
+        sep="\n\n",
+        sep2="<｜end▁of▁sentence｜>",
+        stop_token_ids=[100001],
+        stop_str=["User:", "<｜end▁of▁sentence｜>"],
+    )
+)
+register_conv_template(
+    Conversation(
+        name="deepseek",
+        system_template="{system_message}",
+        # system_message="You are a helpful assistant. Please answer truthfully and write out your "
+        # "thinking step by step to be sure you get the right answer.",
+        system_message="",
+        roles=("<|User|>", "<|Assistant|>"),
+        messages=(),
+        offset=0,
+        sep_style=SeparatorStyle.DeepSeek,
+        sep="\n\n",
+        sep2="<｜end▁of▁sentence｜>",
+        stop_token_ids=[100001],
+        stop_str=["<|User|>", "<｜end▁of▁sentence｜>"]
+    )
+)
+
+register_conv_template(
+    Conversation(
+        name="plain",
+        system_template="",
+        system_message="",
+        roles=("", ""),
+        messages=(),
+        offset=0,
+        sep_style=SeparatorStyle.PLAIN,
+        sep="",
+        sep2="",
+        stop_token_ids=[2],
+        stop_str=["</s>"],
+    )
+)
+
+
+register_conv_template(
+    Conversation(
+        name="alignment",
+        system_template="",
+        system_message="",
+        roles=("", ""),
+        messages=(),
+        offset=0,
+        sep_style=SeparatorStyle.ALIGNMENT,
+        sep="",
+        sep2="",
+        stop_token_ids=[2],
+        stop_str=["</s>"],
+    )
+)
+
+
+if __name__ == "__main__":
+    # print("Llama-2 template:")
+    # conv = get_conv_template("llama-2")
+    # conv.set_system_message("You are a helpful, respectful and honest assistant.")
+    # conv.append_message(conv.roles[0], "Hello!")
+    # conv.append_message(conv.roles[1], "Hi!")
+    # conv.append_message(conv.roles[0], "How are you?")
+    # conv.append_message(conv.roles[1], None)
+    # print(conv.get_prompt())
+
+    # print("\n")
+
+    print("deepseek template:")
+    conv = get_conv_template("deepseek")
+    conv.append_message(conv.roles[0], "Hello!")
+    conv.append_message(conv.roles[1], "Hi! This is Tony.")
+    conv.append_message(conv.roles[0], "Who are you?")
+    conv.append_message(conv.roles[1], "I am a helpful assistant.")
+    conv.append_message(conv.roles[0], "How are you?")
+    conv.append_message(conv.roles[1], None)
+    print(conv.get_prompt())

tokenizer_config.json

@@ -0,0 +1,11 @@
+{
+  "bos_token": "<｜begin▁of▁sentence｜>",
+  "clean_up_tokenization_spaces": false,
+  "eos_token": "<｜end▁of▁sentence｜>",
+  "model_max_length": 16384,
+  "pad_token": null,
+  "tokenizer_class": "LlamaTokenizer",
+  "chat_template": "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% set ns = namespace(is_first=false, is_tool=false, is_output_first=true, system_prompt='') %}{%- for message in messages %}{%- if message['role'] == 'system' %}{% set ns.system_prompt = message['content'] %}{%- endif %}{%- endfor %}{{ns.system_prompt}}{%- for message in messages %}{%- if message['role'] == 'user' %}{{'<|User|>: ' + message['content'] + '\\n\\n'}}{%- endif %}{%- if message['role'] == 'assistant' and message['content'] is not none %}{% set content = message['content'] %}{% if '</think>' in content %}{% set content = content.split('</think>')[-1] %}{% endif %}{{'<|Assistant|>: ' + content + '<｜end▁of▁sentence｜>'}}{%- endif %}{%- endfor -%}{% if add_generation_prompt %}{{'<|Assistant|>: '}}{% endif %}",
+  "unk_token": null,
+  "use_default_system_prompt": true
+}