fix `_name_or_path` in config.json

README.md

@@ -40,6 +40,12 @@ The training setup was `4xA100's 80GB` and took ~6 hours to pretrain and ~13 hou
 ---
 
 
+## Installing requirements
+```
+pip install -qr https://huggingface.co/Tensoic/Cerule-v0.1/resolve/main/requirements.txt
+```
+
+
 ## Training:
 We will release the training code in some time.
 

config.json

@@ -1,5 +1,5 @@
 {
- "_name_or_path": "Tensoic/Cerule",
+ "_name_or_path": "Tensoic/Cerule-v0.1",
  "architectures": [
  "CeruleGemmaForCausalLM"
  ],
@@ -34,7 +34,7 @@
  "tokenizer_model_max_length": 2048,
  "tokenizer_padding_side": "right",
  "torch_dtype": "bfloat16",
- "transformers_version": "4.39.0.dev0",
+ "transformers_version": "4.39.1",
  "tune_mm_mlp_adapter": false,
  "use_cache": true,
  "use_mm_proj": true,

requirements.txt

@@ -0,0 +1,2 @@
+transformers>=4.39.1 
+flash_attn