Upload folder using huggingface_hub

config.json

@@ -40,13 +40,15 @@
   "quantization_config": {
     "_load_in_4bit": true,
     "_load_in_8bit": false,
-    "bnb_4bit_compute_dtype": "bfloat16",
-    "bnb_4bit_quant_storage": "float32",
-    "bnb_4bit_quant_type": "nf4",
-    "bnb_4bit_use_double_quant": true,
+    "bnb_4bit_compute_dtype": "float32",
+    "bnb_4bit_quant_storage": "uint8",
+    "bnb_4bit_quant_type": "fp4",
+    "bnb_4bit_use_double_quant": false,
     "llm_int8_enable_fp32_cpu_offload": false,
     "llm_int8_has_fp16_weight": false,
-    "llm_int8_skip_modules": null,
+    "llm_int8_skip_modules": [
+      "mamba"
+    ],
     "llm_int8_threshold": 6.0,
     "load_in_4bit": true,
     "load_in_8bit": false,
@@ -56,9 +58,9 @@
   "router_aux_loss_coef": 0.001,
   "sliding_window": null,
   "tie_word_embeddings": false,
-  "torch_dtype": "bfloat16",
+  "torch_dtype": "float16",
   "transformers_version": "4.40.0.dev0",
-  "use_cache": false,
+  "use_cache": true,
   "use_mamba_kernels": true,
   "vocab_size": 65536
 }

model-00001-of-00007.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:1a439b2047a16c590b63a7a8f759ea4bb7a8f5205eb07f8ed2f83d39b71a2f55
+oid sha256:2e92c801b498d391703c2d2aa7ecfd5251c43cb8fcea5469a35cba13e6e91fb8
 size 4873003324

model-00002-of-00007.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:d9769f0709625f7c4596eb1be8e2baacfd8f388092746d4406b725a8e577144e
+oid sha256:b5931dc5bfcf2db3a287940e23861ffb9a9661e76e824cedb0a73d8cc359c2f0
 size 4984623243

model-00003-of-00007.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:e2a9537e3542d8016fd4daeab9b3ef4058ceefdc35f24ed3fc3ace13948310c4
+oid sha256:ec106651959c1fc61482e954d6c1dee67d82d387f8c7edf366a1b89f9d14e027
 size 4983413752

model-00004-of-00007.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f1c6103bbf6db840d25d6748ee80d7df09d0fbd0e94ebeb57bb9806eb8f4715f
+oid sha256:5f5178778f3ee8501126fe41ba37b8d7fd3837fcc5b2e902779ddba7e801ed42
 size 4997975503

model-00005-of-00007.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:afb102dd8aac845bba0f9d495ca5c8095a1f6be80c22f65cb08bc295a23680aa
+oid sha256:ea765123548c90cc441e70a0c8f6d766208994730bf03f41c6b4f18be1f5ed9a
 size 4983413840

model-00006-of-00007.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c0e9210edb8e25246c3eb7ff9bd7f27edefc012c96e35027352d0a07eeef41b7
+oid sha256:f25d10f17ca3965392e5c876a59d901d89a18bed479e8fa790c375ac3b40a9a3
 size 4975844741

model-00007-of-00007.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:aaaae336c946d1f5ab14a699c1543b424533f718895dc5305638e55329802df6
+oid sha256:24d0e9a3fb0a6f8c61489778097d6debaf5da9ccc77299d1fef5fbb19b02ae9c
 size 3842901815

tokenizer_config.json

@@ -36,7 +36,6 @@
     }
   },
   "bos_token": "<|startoftext|>",
-  "chat_template": "{% if not add_generation_prompt is defined %}{% set add_generation_prompt = false %}{% endif %}{% for message in messages %}{{'<|im_start|>' + message['role'] + '\n' + message['content'] + '<|im_end|>' + '\n'}}{% endfor %}{% if add_generation_prompt %}{{ '<|im_start|>assistant\n' }}{% endif %}",
   "clean_up_tokenization_spaces": false,
   "eos_token": "<|endoftext|>",
   "model_max_length": 1000000000000000019884624838656,