Upload folder using huggingface_hub

README.md

@@ -4,9 +4,9 @@ tags:
 - mlx
 ---
 
-# GreenBitAI/01-Yi-6B-chat-layer-mix-bpw-2.2-mlx
-This quantized low-bit model was converted to MLX format from [`GreenBitAI/01-Yi-6B-chat-layer-mix-bpw-2.2`]().
-Refer to the [original model card](https://huggingface.co/GreenBitAI/01-Yi-6B-chat-layer-mix-bpw-2.2) for more details on the model.
+# GreenBitAI/01-Yi-6B-Chat-layer-mix-bpw-2.2-mlx
+This quantized low-bit model was converted to MLX format from [`GreenBitAI/01-Yi-6B-Chat-layer-mix-bpw-2.2`]().
+Refer to the [original model card](https://huggingface.co/GreenBitAI/01-Yi-6B-Chat-layer-mix-bpw-2.2) for more details on the model.
 ## Use with mlx
 
 ```bash
@@ -16,6 +16,6 @@ pip install gbx-lm
 ```python
 from gbx_lm import load, generate
 
-model, tokenizer = load("GreenBitAI/01-Yi-6B-chat-layer-mix-bpw-2.2-mlx")
+model, tokenizer = load("GreenBitAI/01-Yi-6B-Chat-layer-mix-bpw-2.2-mlx")
 response = generate(model, tokenizer, prompt="hello", verbose=True)
 ```

config.json

@@ -77,7 +77,7 @@
     "top_p": 1.0,
     "torch_dtype": "float16",
     "torchscript": false,
-    "transformers_version": "4.39.0.dev0",
+    "transformers_version": "4.39.3",
     "typical_p": 1.0,
     "use_bfloat16": false,
     "use_cache": true,

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8719d9812ae0a98010747e74769fc95f53fa5b2c17b3074219f99a1ce310dc20
-size 2868359625
+oid sha256:5c7c8e4b071f59e95c8ca9fbc739394c897f070788dba38473bcc06f52591a6d
+size 2867900553

model.safetensors.index.json

@@ -1,6 +1,6 @@
 {
     "metadata": {
-        "total_size": 2868224000
+        "total_size": 2867765248
     },
     "weight_map": {
         "lm_head.weight": "model.safetensors",

quant_strategy.json

@@ -89,8 +89,8 @@
             }
         },
         "model.layers.1": {
-            "accuracy": 0.9216856956481934,
-            "total_bits": 486917664,
+            "accuracy": 0.7773141860961914,
+            "total_bits": 365124672,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -129,10 +129,10 @@
             },
             "o_proj": {
                 "group_size": {
-                    "4": 128
+                    "2": 64
                 },
                 "bits": [
-                    4
+                    2
                 ],
                 "bits_prop": [
                     1
@@ -165,10 +165,10 @@
             },
             "down_proj": {
                 "group_size": {
-                    "4": 128
+                    "2": 64
                 },
                 "bits": [
-                    4
+                    2
                 ],
                 "bits_prop": [
                     1
@@ -265,8 +265,8 @@
             }
         },
         "model.layers.3": {
-            "accuracy": 0.9084997177124023,
-            "total_bits": 449761824,
+            "accuracy": 0.8372340202331543,
+            "total_bits": 360997440,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -341,10 +341,10 @@
             },
             "down_proj": {
                 "group_size": {
-                    "4": 128
+                    "2": 64
                 },
                 "bits": [
-                    4
+                    2
                 ],
                 "bits_prop": [
                     1
@@ -705,8 +705,8 @@
             }
         },
         "model.layers.8": {
-            "accuracy": 0.9228010177612305,
-            "total_bits": 449761824,
+            "accuracy": 0.7149877548217773,
+            "total_bits": 360997440,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -781,10 +781,10 @@
             },
             "down_proj": {
                 "group_size": {
-                    "4": 128
+                    "2": 64
                 },
                 "bits": [
-                    4
+                    2
                 ],
                 "bits_prop": [
                     1
@@ -793,8 +793,8 @@
             }
         },
         "model.layers.9": {
-            "accuracy": 0.9557886123657227,
-            "total_bits": 360997440,
+            "accuracy": 0.9577234387397766,
+            "total_bits": 365124672,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -809,10 +809,10 @@
             },
             "k_proj": {
                 "group_size": {
-                    "2": 64
+                    "4": 128
                 },
                 "bits": [
-                    2
+                    4
                 ],
                 "bits_prop": [
                     1
@@ -1057,8 +1057,8 @@
             }
         },
         "model.layers.12": {
-            "accuracy": 0.9378215074539185,
-            "total_bits": 360997440,
+            "accuracy": 0.9404451847076416,
+            "total_bits": 365124672,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -1073,10 +1073,10 @@
             },
             "k_proj": {
                 "group_size": {
-                    "2": 64
+                    "4": 128
                 },
                 "bits": [
-                    2
+                    4
                 ],
                 "bits_prop": [
                     1
@@ -1233,8 +1233,8 @@
             }
         },
         "model.layers.14": {
-            "accuracy": 0.9317333698272705,
-            "total_bits": 360997440,
+            "accuracy": 0.9359749555587769,
+            "total_bits": 365124672,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -1249,10 +1249,10 @@
             },
             "k_proj": {
                 "group_size": {
-                    "2": 64
+                    "4": 128
                 },
                 "bits": [
-                    2
+                    4
                 ],
                 "bits_prop": [
                     1
@@ -1761,8 +1761,8 @@
             }
         },
         "model.layers.20": {
-            "accuracy": 0.9032342433929443,
-            "total_bits": 360997440,
+            "accuracy": 0.9059760570526123,
+            "total_bits": 365124672,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -1777,10 +1777,10 @@
             },
             "k_proj": {
                 "group_size": {
-                    "2": 64
+                    "4": 128
                 },
                 "bits": [
-                    2
+                    4
                 ],
                 "bits_prop": [
                     1
@@ -1849,8 +1849,8 @@
             }
         },
         "model.layers.21": {
-            "accuracy": 0.9000318050384521,
-            "total_bits": 360997440,
+            "accuracy": 0.9023456573486328,
+            "total_bits": 365124672,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -1865,10 +1865,10 @@
             },
             "k_proj": {
                 "group_size": {
-                    "2": 64
+                    "4": 128
                 },
                 "bits": [
-                    2
+                    4
                 ],
                 "bits_prop": [
                     1
@@ -2289,8 +2289,8 @@
             }
         },
         "model.layers.26": {
-            "accuracy": 0.9015021324157715,
-            "total_bits": 360997440,
+            "accuracy": 0.9034652709960938,
+            "total_bits": 365124672,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -2305,10 +2305,10 @@
             },
             "k_proj": {
                 "group_size": {
-                    "2": 64
+                    "4": 128
                 },
                 "bits": [
-                    2
+                    4
                 ],
                 "bits_prop": [
                     1
@@ -2465,8 +2465,8 @@
             }
         },
         "model.layers.28": {
-            "accuracy": 0.9067087173461914,
-            "total_bits": 360997440,
+            "accuracy": 0.9271254539489746,
+            "total_bits": 449761824,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -2541,10 +2541,10 @@
             },
             "down_proj": {
                 "group_size": {
-                    "2": 64
+                    "4": 128
                 },
                 "bits": [
-                    2
+                    4
                 ],
                 "bits_prop": [
                     1
@@ -2553,8 +2553,8 @@
             }
         },
         "model.layers.29": {
-            "accuracy": 0.9141254425048828,
-            "total_bits": 360997440,
+            "accuracy": 0.9337887763977051,
+            "total_bits": 449761824,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -2629,10 +2629,10 @@
             },
             "down_proj": {
                 "group_size": {
-                    "2": 64
+                    "4": 128
                 },
                 "bits": [
-                    2
+                    4
                 ],
                 "bits_prop": [
                     1
@@ -2641,8 +2641,8 @@
             }
         },
         "model.layers.30": {
-            "accuracy": 0.9168109893798828,
-            "total_bits": 360997440,
+            "accuracy": 0.9370880126953125,
+            "total_bits": 449761824,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -2717,10 +2717,10 @@
             },
             "down_proj": {
                 "group_size": {
-                    "2": 64
+                    "4": 128
                 },
                 "bits": [
-                    2
+                    4
                 ],
                 "bits_prop": [
                     1
@@ -2729,8 +2729,8 @@
             }
         },
         "model.layers.31": {
-            "accuracy": 0.9416450262069702,
-            "total_bits": 660324384,
+            "accuracy": 0.9427725076675415,
+            "total_bits": 664451616,
             "q_proj": {
                 "group_size": {
                     "2": 64
@@ -2745,10 +2745,10 @@
             },
             "k_proj": {
                 "group_size": {
-                    "2": 64
+                    "4": 128
                 },
                 "bits": [
-                    2
+                    4
                 ],
                 "bits_prop": [
                     1

tokenizer.json

@@ -14,20 +14,20 @@
     },
     {
       "id": 1,
-      "content": "<s>",
+      "content": "<|startoftext|>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
-      "normalized": false,
+      "normalized": true,
       "special": true
     },
     {
       "id": 2,
-      "content": "</s>",
+      "content": "<|endoftext|>",
       "single_word": false,
       "lstrip": false,
       "rstrip": false,
-      "normalized": false,
+      "normalized": true,
       "special": true
     },
     {
@@ -56,24 +56,6 @@
       "rstrip": false,
       "normalized": false,
       "special": true
-    },
-    {
-      "id": 64000,
-      "content": "<|startoftext|>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": true
-    },
-    {
-      "id": 64001,
-      "content": "<|endoftext|>",
-      "single_word": false,
-      "lstrip": false,
-      "rstrip": false,
-      "normalized": true,
-      "special": true
     }
   ],
   "normalizer": {
@@ -153,8 +135,8 @@
     "byte_fallback": true,
     "vocab": {
       "<unk>": 0,
-      "<s>": 1,
-      "</s>": 2,
+      "<|startoftext|>": 1,
+      "<|endoftext|>": 2,
       "<|Human|>": 3,
       "<|Assistant|>": 4,
       "<|System|>": 5,

tokenizer_config.json

@@ -12,17 +12,17 @@
       "special": true
     },
     "1": {
-      "content": "<s>",
+      "content": "<|startoftext|>",
       "lstrip": false,
-      "normalized": false,
+      "normalized": true,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
     "2": {
-      "content": "</s>",
+      "content": "<|endoftext|>",
       "lstrip": false,
-      "normalized": false,
+      "normalized": true,
       "rstrip": false,
       "single_word": false,
       "special": true
@@ -50,22 +50,6 @@
       "rstrip": false,
       "single_word": false,
       "special": true
-    },
-    "64000": {
-      "content": "<|startoftext|>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": true
-    },
-    "64001": {
-      "content": "<|endoftext|>",
-      "lstrip": false,
-      "normalized": true,
-      "rstrip": false,
-      "single_word": false,
-      "special": true
     }
   },
   "additional_special_tokens": [