Hugging Face's logo

wanderer2k1
/
base_LLM_183MB_concat_and_chunk

TensorBoard
Safetensors
mpt
Model card Files Metrics Community
wanderer2k1 commited on
Commit
bf5f3e5
•
1 Parent(s): 8686ab2

Training in progress, step 200

Browse files
Files changed (9) hide show
  1. config.json +33 -0
  2. model.safetensors +3 -0
  3. runs/Sep24_01-37-36_kao-dgxa-e10-u17/events.out.tfevents.1727141863.kao-dgxa-e10-u17.1357.0 +3 -0
  4. runs/Sep24_02-04-50_kao-dgxa-e10-u17/events.out.tfevents.1727143510.kao-dgxa-e10-u17.1357.1 +3 -0
  5. runs/Sep24_02-25-27_kao-dgxa-e10-u17/events.out.tfevents.1727144739.kao-dgxa-e10-u17.2488.0 +3 -0
  6. runs/Sep24_02-25-27_kao-dgxa-e10-u17/events.out.tfevents.1727144809.kao-dgxa-e10-u17.2488.1 +3 -0
  7. runs/Sep24_02-28-38_kao-dgxa-e10-u17/events.out.tfevents.1727144926.kao-dgxa-e10-u17.2963.0 +3 -0
  8. runs/Sep24_02-30-46_kao-dgxa-e10-u17/events.out.tfevents.1727145059.kao-dgxa-e10-u17.1688.0 +3 -0
  9. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,33 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "abili": false,
3
+ "architectures": [
4
+ "MptForCausalLM"
5
+ ],
6
+ "attn_config": {
7
+ "model_type": ""
8
+ },
9
+ "d_model": 768,
10
+ "emb_pdrop": 0.0,
11
+ "embedding_fraction": 1.0,
12
+ "expansion_ratio": 4,
13
+ "init_device": "cpu",
14
+ "initializer_range": 0.02,
15
+ "layer_norm_epsilon": 1e-05,
16
+ "learned_pos_emb": true,
17
+ "logit_scale": null,
18
+ "max_seq_len": 8192,
19
+ "model_type": "mpt",
20
+ "n_heads": 12,
21
+ "n_layers": 12,
22
+ "no_bias": true,
23
+ "norm_type": "low_precision_layernorm",
24
+ "pad_token_id": 128009,
25
+ "resid_pdrop": 0.0,
26
+ "rope": true,
27
+ "rope_theta": 100000,
28
+ "torch_dtype": "float32",
29
+ "transformers_version": "4.40.0",
30
+ "use_cache": false,
31
+ "verbose": 0,
32
+ "vocab_size": 128256
33
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:62df07e31ba0c73a5a56b2a4ea9d826341c43f171c23d706cc7b57885e3cb6b0
3
+ size 733826032
runs/Sep24_01-37-36_kao-dgxa-e10-u17/events.out.tfevents.1727141863.kao-dgxa-e10-u17.1357.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:127263177b86e482f3a4b088bdd31017066a3601b5e3cf19031c0f458bef4968
3
+ size 104194
runs/Sep24_02-04-50_kao-dgxa-e10-u17/events.out.tfevents.1727143510.kao-dgxa-e10-u17.1357.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e56c1f1d5a963f8ce9e0858c3c6e3bbcdbbf6b25f18246f6961d0ab91a77c7c8
3
+ size 12099
runs/Sep24_02-25-27_kao-dgxa-e10-u17/events.out.tfevents.1727144739.kao-dgxa-e10-u17.2488.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:642e2bdaaf627a17ebbe83b6c492ac42c8115bc0c3b48da522776410512beaec
3
+ size 4771
runs/Sep24_02-25-27_kao-dgxa-e10-u17/events.out.tfevents.1727144809.kao-dgxa-e10-u17.2488.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b1c27542fa232eea11460683891f2678ddc2897438ea0a7ca7d1a72ded5ce752
3
+ size 7876
runs/Sep24_02-28-38_kao-dgxa-e10-u17/events.out.tfevents.1727144926.kao-dgxa-e10-u17.2963.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f1dc7cb64e3bb78bd2f1320157eaec3e9e1294ba0bd92a20ed8ff2fdd41f1297
3
+ size 4979
runs/Sep24_02-30-46_kao-dgxa-e10-u17/events.out.tfevents.1727145059.kao-dgxa-e10-u17.1688.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a2a98b37a2dc4759925c85a416fbd0df3df4c835c3b06e56a99760decd2c04cb
3
+ size 5607
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:dde9191dd71f7bd05f5aac9c7a9d24440d1bc97930e725db2fb8932c361225a0
3
+ size 5176