Add files using upload-large-folder tool

config.json

@@ -18,6 +18,7 @@
   "num_attention_heads": 32,
   "num_hidden_layers": 16,
   "num_key_value_heads": 8,
+  "pad_token_id": 128004,
   "pretraining_tp": 1,
   "quantization_config": {
     "_load_in_4bit": true,
@@ -28,7 +29,12 @@
     "bnb_4bit_use_double_quant": true,
     "llm_int8_enable_fp32_cpu_offload": false,
     "llm_int8_has_fp16_weight": false,
-    "llm_int8_skip_modules": null,
+    "llm_int8_skip_modules": [
+      "lm_head",
+      "multi_modal_projector",
+      "merger",
+      "modality_projection"
+    ],
     "llm_int8_threshold": 6.0,
     "load_in_4bit": true,
     "load_in_8bit": false,
@@ -45,7 +51,8 @@
   "rope_theta": 500000.0,
   "tie_word_embeddings": true,
   "torch_dtype": "bfloat16",
-  "transformers_version": "4.44.2",
+  "transformers_version": "4.48.1",
+  "unsloth_fixed": true,
   "use_cache": true,
   "vocab_size": 128256
 }

generation_config.json

@@ -1,6 +1,11 @@
 {
   "_from_model_config": true,
   "bos_token_id": 128000,
+  "do_sample": true,
   "eos_token_id": 128001,
-  "transformers_version": "4.44.2"
+  "max_length": 131072,
+  "pad_token_id": 128004,
+  "temperature": 0.6,
+  "top_p": 0.9,
+  "transformers_version": "4.48.1"
 }

tokenizer.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:79e3e522635f3171300913bb421464a87de6222182a0570b9b2ccba2a964b2b4
-size 9085657
+oid sha256:6b9e4e7fb171f92fd137b777cc2714bf87d11576700a1dcd7a399e7bbe39537b
+size 17209920

tokenizer_config.json

@@ -1,4 +1,5 @@
 {
+  "add_bos_token": true,
   "added_tokens_decoder": {
     "128000": {
       "content": "<|begin_of_text|>",
@@ -2052,6 +2053,7 @@
   "bos_token": "<|begin_of_text|>",
   "clean_up_tokenization_spaces": true,
   "eos_token": "<|end_of_text|>",
+  "extra_special_tokens": {},
   "model_input_names": [
     "input_ids",
     "attention_mask"
@@ -2059,5 +2061,6 @@
   "model_max_length": 131072,
   "pad_token": "<|finetune_right_pad_id|>",
   "padding_side": "left",
-  "tokenizer_class": "PreTrainedTokenizerFast"
+  "tokenizer_class": "PreTrainedTokenizerFast",
+  "unk_token": null
 }