Hugging Face's logo

dhanesh123in
/
HGRNBitCausalLM

Text Generation
Transformers
TensorBoard
Safetensors
hgrn_bit
Inference Endpoints
Model card Files Metrics Community
dhanesh123in commited on
Commit
cac9216
•
1 Parent(s): 75aeb69

Training in progress, step 500

Browse files
Files changed (16) hide show
  1. config.json +30 -0
  2. model.safetensors +3 -0
  3. runs/Jul03_22-07-38_morpheus/events.out.tfevents.1720024661.morpheus.731446.1 +3 -0
  4. runs/Jul03_23-02-04_morpheus/events.out.tfevents.1720027927.morpheus.748838.0 +3 -0
  5. runs/Jul03_23-33-54_morpheus/events.out.tfevents.1720029834.morpheus.761495.0 +3 -0
  6. runs/Jul03_23-42-30_morpheus/events.out.tfevents.1720030352.morpheus.761495.1 +3 -0
  7. runs/Jul04_00-09-56_morpheus/events.out.tfevents.1720031996.morpheus.770210.0 +3 -0
  8. runs/Jul04_00-18-49_morpheus/events.out.tfevents.1720032530.morpheus.770210.1 +3 -0
  9. runs/Jul04_00-23-10_morpheus/events.out.tfevents.1720032791.morpheus.770210.2 +3 -0
  10. runs/Jul04_00-25-32_morpheus/events.out.tfevents.1720032933.morpheus.774214.0 +3 -0
  11. runs/Jul04_00-31-29_morpheus/events.out.tfevents.1720033290.morpheus.774214.1 +3 -0
  12. runs/Jul04_00-32-06_morpheus/events.out.tfevents.1720033327.morpheus.774214.2 +3 -0
  13. runs/Jul04_00-50-50_morpheus/events.out.tfevents.1720034451.morpheus.774214.3 +3 -0
  14. runs/Jul04_00-52-46_morpheus/events.out.tfevents.1720034567.morpheus.774214.4 +3 -0
  15. runs/Jul04_00-58-37_morpheus/events.out.tfevents.1720034917.morpheus.774214.5 +3 -0
  16. training_args.bin +3 -0
config.json ADDED
@@ -0,0 +1,30 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "_name_or_path": "ridger/MMfreeLM-370M",
3
+ "architectures": [
4
+ "HGRNBitForCausalLM"
5
+ ],
6
+ "attn_mode": "fused_recurrent",
7
+ "bos_token_id": 1,
8
+ "conv_size": 4,
9
+ "eos_token_id": 2,
10
+ "expand_ratio": 1,
11
+ "fuse_cross_entropy": true,
12
+ "hidden_act": "swish",
13
+ "hidden_ratio": 4,
14
+ "hidden_size": 1024,
15
+ "initializer_range": 0.02,
16
+ "intermediate_size": null,
17
+ "max_position_embeddings": 2048,
18
+ "model_type": "hgrn_bit",
19
+ "num_heads": 1,
20
+ "num_hidden_layers": 24,
21
+ "rms_norm_eps": 1e-06,
22
+ "share_conv_kernel": true,
23
+ "tie_word_embeddings": false,
24
+ "torch_dtype": "float32",
25
+ "transformers_version": "4.42.3",
26
+ "use_cache": true,
27
+ "use_lower_bound": true,
28
+ "use_short_conv": false,
29
+ "vocab_size": 32000
30
+ }
model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:daf25cf64d74651e362aa78acae6d16a31d86a61f5597036a6f30560d62c8a46
3
+ size 1496472568
runs/Jul03_22-07-38_morpheus/events.out.tfevents.1720024661.morpheus.731446.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:07e745b9f0667c0a12cb503067c3f46445ecbf6e4843c698aec34c7a5a902d3f
3
+ size 4785
runs/Jul03_23-02-04_morpheus/events.out.tfevents.1720027927.morpheus.748838.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:0aeb8ba2bccf022c9207633499e014161fc7ef4ffdc197de43f20708457d90a2
3
+ size 4908
runs/Jul03_23-33-54_morpheus/events.out.tfevents.1720029834.morpheus.761495.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2b645d102823a020fd9d2f0b1656ed0d3ba9c02f2c329a0003533da09efc858c
3
+ size 4908
runs/Jul03_23-42-30_morpheus/events.out.tfevents.1720030352.morpheus.761495.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:58b455842baa3837ae4742ba62467c834b268cf3d432f6aacdaa8cf0274a36d1
3
+ size 4828
runs/Jul04_00-09-56_morpheus/events.out.tfevents.1720031996.morpheus.770210.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a59fad71ad8df7eb161fc19b2f3487938b43a434b436460ef4b0bc78bcad074b
3
+ size 4828
runs/Jul04_00-18-49_morpheus/events.out.tfevents.1720032530.morpheus.770210.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ee441bf55244aaf2e8b2c65eae92296d130754cd5dda0c7f60cc8ceb0897c9e8
3
+ size 4828
runs/Jul04_00-23-10_morpheus/events.out.tfevents.1720032791.morpheus.770210.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:52f38651f3b753425180fab04ed999cb81fc2b00848d50c2627a9d5736533c8f
3
+ size 4828
runs/Jul04_00-25-32_morpheus/events.out.tfevents.1720032933.morpheus.774214.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:31d67cdb0ccbf4502689f1abeac6df0647331051eda2ae1e6e4fc20da7c0ae7b
3
+ size 5039
runs/Jul04_00-31-29_morpheus/events.out.tfevents.1720033290.morpheus.774214.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1036e6cf2c96aa2afbd049483f5a3045a0d008f5688e97bd86bb38bdf302eb5a
3
+ size 4827
runs/Jul04_00-32-06_morpheus/events.out.tfevents.1720033327.morpheus.774214.2 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:18d2efaf9dc5e27c03ea43bf7ae19cd8003d5092365e14442b1706caa8bb5eec
3
+ size 5882
runs/Jul04_00-50-50_morpheus/events.out.tfevents.1720034451.morpheus.774214.3 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:77d06c2bd847bd912428847361d7e11663feb00479704c615a5dc986909b7c07
3
+ size 4826
runs/Jul04_00-52-46_morpheus/events.out.tfevents.1720034567.morpheus.774214.4 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6349e0e60b7ca25615dc90d76b0373617055c2efcff679bf94c5fe2007cd377c
3
+ size 5037
runs/Jul04_00-58-37_morpheus/events.out.tfevents.1720034917.morpheus.774214.5 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1b7a10987f6db41d1ec1da22156919c3bac9df3698fff4f6bbb097ea4d35afbc
3
+ size 5308
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c1870d4d51db1b026ecf71368b6c3f0fe23d108d63d80f44367042872dbac536
3
+ size 5112