Hugging Face's logo

stojchet
/
sft1

Text Generation
Transformers
Safetensors
llama
trl
sft
Generated from Trainer
text-generation-inference
Inference Endpoints
Model card Files Community
stojchet commited on
Commit
dd9296c
·
verified ·
1 Parent(s): 9ff8fa7

Training in progress, step 129

Browse files
Files changed (5) hide show
  1. config.json +1 -2
  2. model-00001-of-00002.safetensors +1 -1
  3. model-00002-of-00002.safetensors +1 -1
  4. tokenizer_config.json +1 -1
  5. training_args.bin +1 -1
config.json CHANGED
@@ -1,5 +1,4 @@
1
  {
2
- "_name_or_path": "deepseek-ai/deepseek-coder-1.3b-base",
3
  "architectures": [
4
  "LlamaForCausalLM"
5
  ],
@@ -28,7 +27,7 @@
28
  "rope_theta": 100000,
29
  "tie_word_embeddings": false,
30
  "torch_dtype": "float32",
31
- "transformers_version": "4.47.0.dev0",
32
  "use_cache": false,
33
  "vocab_size": 32256
34
  }
 
1
  {
 
2
  "architectures": [
3
  "LlamaForCausalLM"
4
  ],
 
27
  "rope_theta": 100000,
28
  "tie_word_embeddings": false,
29
  "torch_dtype": "float32",
30
+ "transformers_version": "4.50.0.dev0",
31
  "use_cache": false,
32
  "vocab_size": 32256
33
  }
model-00001-of-00002.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:243061629ae1f64a33ac5a15ee6d88350b1e33a24263a2996113130fe1367afc
3
  size 4986380064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e5708d730bc949f6395c08ec1f8307312c6908052a740904c96bcd487c99e3e
3
  size 4986380064
model-00002-of-00002.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e9d4c9bb2f21b1ccb7bdc0f4e2d2c9380f1b494c315833dd41be31ba0ce0a754
3
  size 399532808
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c96eefd3b0fbf894b46e8cb4e48885086df304e57417c1127b64edfeceee01de
3
  size 399532808
tokenizer_config.json CHANGED
@@ -188,7 +188,7 @@
188
  "model_max_length": 16384,
189
  "pad_token": "<｜end▁of▁sentence｜>",
190
  "sp_model_kwargs": {},
191
- "tokenizer_class": "LlamaTokenizer",
192
  "unk_token": null,
193
  "use_default_system_prompt": false
194
  }
 
188
  "model_max_length": 16384,
189
  "pad_token": "<｜end▁of▁sentence｜>",
190
  "sp_model_kwargs": {},
191
+ "tokenizer_class": "LlamaTokenizerFast",
192
  "unk_token": null,
193
  "use_default_system_prompt": false
194
  }
training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fb5964f02119013445810ab046d46126efa3fc6891c9c6fdc4f60cc78d8aa0db
3
  size 5368
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7d26b52002bd08704fc9764f6dca17299bd82bb197c9261318a441182505929f
3
  size 5368